What Comes Out, Must Go In: Macronutrient balance assessment of transitioning home garden systems in southern Ethiopia

Nadine Galle (10850155)

Dr. L.H. Cammeraat Dr. G.W.J. van de Ven Dr. K.K.E. Descheemaeker Drs. B.T. Mellisse Dr. B. Jansen

Master Earth Science - University of Amsterdam Institute for Biodiversity and Ecosystem Dynamics

MSc. Thesis | UvA 5264MTR30Y | WUR PPS-80430 Environmental Management | 30 ECTS October 2015 - April 2016 April 1, 2016

pg. 1

MAIN APPLICANT

Nadine Galle Schoolmeesterstraat 24 1053 MC Amsterdam The Netherlands

+ 31 6 51 55 48 18 [email protected]

University of Amsterdam Institute for Biodiversity and Ecosystem Dynamics (IBED) Amsterdam, The Netherlands 5264MTR30Y Master Thesis Earth Sciences Environmental Management Track Student No. 10850155

Supervisor: Dr. L.H. (Erik) Cammeraat Second reader: Dr. B. (Boris) Jansen

Wageningen University and Research Center Plant Production Systems (PPS) Wageningen, The Netherlands PPS-80430 Master Thesis Plant Production Systems Registration No. 920610-249-130

Supervisor: Dr. G.W.J. (Gerrie) van de Ven Co-Supervisor: Dr. K.K.E. (Katrien) Descheemaeker

Hawassa University Hawassa University College of Agriculture Wondo Genet College of Forestry and Natural Resources Hawassa, Sidama, Ethiopia

Daily supervisor: Drs. B.T. (Beyene) Mellisse

I declare that the work I am submitting for assessment contains no section copied in whole or in part from any other source unless explicitly identified in quotation marks and with detailed, complete and accurate referencing.

Signed, Nadine J. Galle

pg. 2

The cattle is only as good as the pasture in which it grazes.

Ethiopian Proverb

pg. 3

pg. 4

ABSTRACT

What Comes Out, Must Go in: Macronutrient balance assessment of transitioning home gardens in southern Ethiopia

By Nadine Galle

Smallholder-operated home garden agroforestry systems are the backbone of Ethiopia’s agricultural sector. In southern Ethiopia, the enset (Enset ventricosum) and coffee (Coffea arabica) based home gardens have sustained millions of livelihoods for centuries, combining subsistence agriculture with a small cash crop income. Enset withstands drought, produces large volumes of food per unit area and is exclusively fertilized with organic matter, an internal input. The resilience of these systems relies on efficient nutrient cycling and multi-species composition. However, population growth induced land fragmentation has led to rapid replacement of enset and coffee with khat (Catha edulis), a lucrative cash crop and popular stimulant. Khat has expanded at the expense of land allocated to enset and coffee and threatens well-established internal nutrient flows within home gardens. The transition called for the definition of five distinct home garden types: four enset-oriented (enset-based, enset- coffee, enset-cereal-vegetable, and enset-livestock) and one khat-based. This paper describes macronutrient (NPK) balances calculated at component and farm level in Sidama and Gedeo, southern Ethiopia. Fields with the same or similar crop were grouped into five farm ‘components’. Livestock was also a component. Representative farms for each home garden type were conceived based on component land use. Processes quantified included mineral fertilizer, organic matter, internal fodder, external fodder and harvested products, removed crop residues, household livestock consumption, harvested products sold off-farm and whole livestock and livestock products sold off-farm. Component level balances added valued to the study by permitting comparison of internal flows, demonstrating the inherent diversity and complexity within home garden systems. Nutrient balances at the farm level showed positive nitrogen (N) balances, fluctuating phosphorus (P) balances and deficient potassium (K) balances, amongst all representative farms. Component level balances were similar but revealed the most severe K deficiencies were in the khat component. Measurements to address nutrient deficiencies, such as enset leaves as crop residue and proper manure handling, were presented and the urgency to develop strategies to reverse khat expansion at the expense of enset was stressed.

Keywords: nutrient balance, nutrient management, nitrogen, phosphorus, potassium, home garden, agroforestry, Ensete ventricosum (enset), Coffea arabica (coffee), Catha edulis (khat)

pg. 5

pg. 6

TABLE OF CONTENTS

MAIN APPLICANT 2 ABSTRACT 5 COVER PHOTO 10 PROJECT TITLE 11 GLOSSARY 11 ACRONYMS AND ABBREVIATIONS 12 LIST OF TABLES, FIGURES AND EQUATIONS 13 ACKNOWLEDGEMENTS 17

1. INTRODUCTION 18

1.1 RESEARCH WITHIN THE CASCAPE PROJECT 19

1.2 SOCIETAL AND SCIENTIFIC SIGNIFICANCE 20

1.3 OUTLINE OF THE THESIS 21

2. RESEARCH OBJECTIVES AND QUESTIONS 22 3. THEORETICAL FRAMEWORK 23

3.1 DEFINITION OF CONCEPTUAL TERMS 23

3.1.1 NUTRIENT BALANCES 23 3.1.2 COMPONENT LEVEL NUTRIENT BALANCE 24 3.1.3 NUTRIENT FLOWS 24

3.2 INFLOWS IN TO THE HOME GARDEN SYSTEM 26

3.2.1 MINERAL FERTILIZER (IN1) 26 3.2.2 EXTERNAL LIVESTOCK FODDER (IN4) 27

3.3 OUTFLOWS FROM THE HOME GARDEN 28

3.3.1 REMOVAL IN HARVESTED PRODUCTS SOLD OFF-FARM (OUT5) 28 3.3.2 LIVESTOCK OUTPUT (OUT3) 28

3.4 INTERNAL FLOWS IN THE HOME GARDEN SYSTEM 28

3.4.1 ORGANIC MATTER (IN2) 28 3.4.2 INTERNAL LIVESTOCK FODDER (IN3) 29 3.4.3 REMOVAL IN ALL HARVESTED PRODUCTS (OUT1) 29 3.4.4 REMOVAL IN CROP RESIDUES (OUT2) 30 3.4.5 HOUSEHOLD LIVESTOCK CONSUMPTION (OUT4) 30

pg. 7

4. METHODOLOGY 31

4.1 STUDY AREA 31

4.2 FARM TYPOLOGIES 34

4.3 DATA COLLECTION 34

4.4 EXPERIMENTAL DESIGN 35

4.4.1 THE REPRESENTATIVE FARM 35 4.4.2 QUANTIFYING NUTRIENT FLOWS 37 4.4.3 MACRONUTRIENT INPUT 38 4.4.4 MACRONUTRIENT OUTPUT 40 4.4.5 THE HARVEST INDEX 41 4.4.6 THE ENSET EXCEPTION 41 4.4.7 COMPONENT LEVEL AND FARM LEVEL MACRONUTRIENT BALANCE 43

4.5 ETHICAL CONSIDERATIONS 45

5. RESULTS 46

5.1 THE REPRESENTATIVE FARMS 46

5.2 FARM SIZE 48

5.3 LIVESTOCK POPULATION 50

5.4 COMPONENT LEVEL NUTRIENT BALANCE ASSESSMENT 51

5.4.1 ENSET-BASED 51 5.4.2 ENSET-COFFEE 54 5.4.3 ENSET-CEREAL-VEGETABLE 57 5.4.4 ENSET-LIVESTOCK 60 5.4.5 KHAT-BASED 63

5.5 FARM LEVEL NUTRIENT BALANCE ASSESSMENT 68

5.6 RESULTS PER HECTARE 72

6. DISCUSSION 75

6.1 UNCERTAINTIES 75

6.2 INTERPRETATION AND DISCUSSION OF RESULTS 77

6.2.1 FARM SIZE 78

pg. 8

6.2.2 FARM LEVEL NUTRIENT BALANCES 78 6.2.3 COMPONENT LEVEL: ENSET 80 6.2.4 COMPONENT LEVEL: COFFEE AND COFFEE + ENSET INTERCROPPING 81 6.2.5 COMPONENT LEVEL: ANNUAL CEREALS AND VEGETABLES 82 6.2.6 COMPONENT LEVEL: KHAT 82 6.2.7 COMPONENT LEVEL: LIVESTOCK 84

6.3 METHODOLOGICAL IMPROVEMENTS AND SUGGESTIONS FOR FURTHER RESEARCH 86

6.4 MANAGEMENT RECOMMENDATIONS 87

6.4.1 ENSET LEAVES AS CROP RESIDUE OR COMPOST ADDITIVE 87 6.4.2 PROPER MANURE HANDLING 88 6.4.3 NUTRIENT-RELATED CONSEQUENCES OF KHAT EXPANSION 89

7. CONCLUSIONS 90 8. REFERENCES 92 7. APPENDICES 99

7.1 CONVERSION TABLE 99

7.2 NUTRIENT CONTENT 100

7.3 SURVEY: INPUTS AND OUTPUTS OF HOMEGARDEN TYPES IN SOUTHERN ETHIOPIA 102

pg. 9

pg. 10

COVER PHOTO

The photograph on the front cover shows two boys amongst their family’s traditional home garden in the Wondo Genet woreda. Photo by Nadine Galle (October 2015).

The photograph on page two illustrates a view of a home garden system in the Malga woreda. Photo by Nadine Galle (December 2015).

PROJECT TITLE

“What Comes Out, Must Go in: Macronutrient balance assessment of transitioning home garden systems in southern Ethiopia”

GLOSSARY

Agroforestry The intentional integrated land use management system, which combines trees and shrubs with crops and/or livestock to create environmental, social and economic benefits. Birr (ETB) The Ethiopian currency. Bula Produced from the inner part of enset and produced into fine powder for high quality pancakes, porridge or dumplings. Charter city A city where the governing system is defined by a city’s charter document, rather than by regional or national laws. In Ethiopia, chartered cities belong to the first level of administrative division (same as kililoch). Fertilizer Any organic or inorganic material of natural or synthetic origin added to soil with the intent to supply one or more plant nutrients essential to growth. Kebele Ethiopia’s fourth and lowest administrative division. Kebeles have similar function to a municipality, neighbourhoods or ward. Kililoch Ethiopia’s first level of administrative division. Since 1995, Ethiopia is constitutionally made up of nine ethically based regional states. The word “kilil” means “reservation” or “protected area”. Kocho Bulk of fermented starch from the enset stem, often made into a pancake from the mixture of scrapped enset sheaths. Woreda Ethiopia’s third level of administrative division. Equivalent to a district. Zone Ethiopia’s second level of administrative division. In Ethiopia, kililoch are further subdivided into 68 zones, these are further divided into woredas. Zurba A bunch of fresh khat leaves, weighing approximately 1 kg.

pg. 11

ACRONYMS AND ABBREVIATIONS

ACV Annual cereals and vegetables AGP Agricultural Growth Programme BNF Biological nitrogen fixation CASCAPE Capacity building for scaling up of evidence-based best practices in agricultural production in Ethiopia CSA Central Statistical Agency of Ethiopia DAP Diammonium phosphate DCM Development of Competitive Markets (Ethiopia) DEP Atmospheric deposition EATA Ethiopia Agricultural Transformation Agency ECI Enset and cereal intercropping ESC Ethiopia Sugar Corporation ETB Ethiopian Birr f1 Nutrient flow: feedstuffs taken from front grazing yard f2 Nutrient flow: cow dung left in front grazing land f3 Nutrient flow: milk and meat consumed by the family f4 Nutrient flow: collection of farm yard manure f5 Nutrient flow: application of farm yard manure in different land use type f6 Nutrient flow: feedstuff taken from different land use type by livestock f7 Nutrient flow: household waste added to manure heap f8 Nutrient flow: family consumption of both perennial and annual crops FGB Farm-Gate Nutrient Balance FYM Farm yard manure GDP Growth Domestic Product GOE Government of Ethiopia GTP Growth and Transformation Plan HwU Hawassa University IBED Institute for Biodiversity and Ecosystem Dynamics IN1 Macronutrient inflow: mineral fertilizer IN2 Macronutrient inflow: organic matter IN3 Macronutrient inflow: internal livestock fodder IN4 Macronutrient inflow: external livestock fodder L1 Losses from front grazing land by leaching, volatilization and erosion L2 Losses from livestock L3 Losses from the manure heap by leaching and volatilization L4 Losses during application of manure to fields L5 Losses from the home garden fields by leaching and volatilization MSA Multivariate Statistical Analysis NPK Nitrogen, Phosphorus and Potassium NUE Nutrient Use Efficiency OUT1 Macronutrient outflow: removal in all harvested products OUT2 Macronutrient outflow: removal in crop residues OUT3 Macronutrient outflow: livestock output OUT4 Macronutrient outflow: household livestock consumption OUT5 Macronutrient outflow: removal in harvested products sold off-farm SNNPR Southern Nations, Nationalities and Peoples’ Region TLU Tropical Livestock Unit TSP Triple superphosphate UvA University of Amsterdam WUR Wageningen University and Research Centre

pg. 12

LIST OF TABLES, FIGURES AND EQUATIONS

Tables

Table 4.1 Agro-ecological zones with characterizing altitude, rainfall, temperature and predominant perennial crops (Mellisse et al., in prep.). Table 4.2 Tropical Livestock Units (TLU) conversion chart (FAO, 1987). Table 4.3 Macronutrient content (mean ±SD) for the four input processes employed in calculating nutrient balances. Table 4.4 Output process and their respective outputs. Table 4.5 Enset output dry matter content (%DM) and nutrient contents (mean ±SD) (HU Agricultural College Soil Lab, 2015; Wondo Genet College Soil Lab, 2015). Table 5.1 Regression equations and R-Squared values for GPS measured land size (ha, x-axis) vs. farmer reported land size (ha, y-axis) by component. Table 5.2 Component level macronutrient inflows (kg/farm/yr) and total sum of nutrient (TSN) from mineral fertilizers (IN1), organic matter (IN2), internal fodder (IN3) and external fodder (IN4) (mean ±SD) by farm component, across five representative farms. Table 5.3 Component level macronutrient outflows (kg/farm/yr) and total sum of nutrient (TSN) from removal in harvested products (OUT1), removal in crop residues (OUT2), whole livestock and livestock products sold off-farm (OUT3) and household livestock consumption (OUT4) (mean ±SD) by farm component, across five representative farms. Table 5.4 Farm level macronutrient inflows (kg/farm/yr) and total sum of nutrient (TSN) from mineral fertilizers (IN1) and external fodder (IN4) (mean ±SD) by farm component, across five representative farms. Table 5.5 Farm level macronutrient outflows (kg/farm/yr) and total sum of nutrient (TSN) from removal in harvested products sold off-farm (OUT5) and whole livestock and livestock products sold off-farm (OUT3) (mean ±SD) by farm component, across five representative farms. Table 5.6 Component level macronutrient inflows, outflows and balances (kg/ha/yr) by farm component, across five representative farms. Table 5.7 Farm level macronutrient inflows, outflows and balances (kg/ha/yr) by farm component, across five representative farms. Table 6.1 Nutrient balance analysis interpretation criteria (expressed as kg of nutrient lost (or added)/ha/yr. Table 6.2 Farm nutrient balances (kg/ha/yr) by representative farm, excluding livestock component. Table 6.3 Farm nutrient balances (kg/ha/yr) for different household groups (Elias et al., 1998; adapted from Roy et al., 2013). Table 6.4 Enset component nutrient balances (kg/ha/yr) by home garden type. Table 6.5 Coffee and enset + coffee intercropping (ECI) component nutrient balances (kg/ha/yr) by home garden type.

pg. 13

Table 6.6 Annual cereal and vegetable (ACV) component nutrient balances (kg/ha/yr) by home garden type. Table 6.7 Khat component nutrient balances (kg/ha/yr) by home garden type. Table 6.8 Livestock component nutrient balances (kg/farm/yr) by home garden type. Table 6.9 Nutrient composition (%) (in dry matter) of manure (Elias et al., 1998), Central Kenyan compost (Lekasi et al., 2003; Kimani & Lekasi, 2004) and compost (this study).

Figures

Figure 1.1 From left to right: 1) the traditional home garden with grazing land in the foreground and behind that, the homestead, then enset infields and coffee/annual cereals and vegetable/khat outfields (Galle, 2015), 2) Enset plants, with Drs. Beyene Mellisse and our translator for scale (Galle, 2015), 3) Women harvesting the enset plant (Mellisse, 2015), 4) Bunches (zurba) of fresh khat leaves (CCTV Africa, 2014). Figure 3.1 Schematic model of nutrient inputs and outputs across the five home garden types, including inputs: atmospheric deposition (DEP), biological nitrogen fixation (BNF), purchased food crops, livestock and farm inputs (Market), cattle which are taken from other farms for fattening purposes (Fat) (e.g. feeding the cattle for three months and then returning them to the owner). The model also shows outputs: losses from front grazing land by leaching, volatilization and erosion (L1), losses from livestock (L2), losses from the manure heap by leaching and volatilization (L3), losses during application of manure to fields (L4), losses from the home garden fields by leaching and volatilization (L5). Nutrient flows on individual home garden include: feedstuffs taken from front grazing yard (f1), cow dung left in front grazing yard (f2) (especially during day time since animals are tied up in grazing yard), milk and meat consumed by the family (f3), collection of farm yard manure (FYM) (f4), application of FYM in different land use type (f5), feedstuff taken from different land use type (especially, enset leaves) by livestock (f6), household waste added to manure heap (f7), family consumption of both perennial and annual crops (f8) (Mellisse et al., in prep.). Figure 4.1 Location of study districts (woredas: Wondo-Genet, Malga, Dale, Bule) within Sidama and Gedeo zones of Southern Nations, Nationalities and Peoples’ Region (SNNPR). The legend displays Ethiopia’s nine regional states and two charted cities. Figure 4.2 Conceptual workflow showing steps and formulae used to extract macronutrient content from input. Figure 4.3 A general nutrient flow diagram of a home garden system. The black dashed line denotes the component level boundary of the farm. Inflows and outflows outside the boundary represent those at the farm level. Thin, gray dashed lines denote relationships excluded from the study. Labels in italic signify factors not quantified, for which it was still worth identifying their place within the system. Figure 5.1 Land use of representative farms expressed in area shares (%).

pg. 14

Figure 5.2 GPS measured land size (ha) vs. farmer reported land size (ha) by farm component. Figure 5.3 Area share of grazing land (ha) in each representative farm vs. average Tropical Livestock Unit (TLU) for each representative farm, by home garden type. Figure 5.4 Component level nutrient inflows, outflows and balances for N, P and K (kg/farm/yr) for an enset-based representative farm. Figure 5.5 Nutrient flows (kg/farm/yr) that influences the nutrient balance of an enset- based system. The asterisk (*) after organic matter (IN2) denotes that this input likely came from an external source, as 0.43 TLU could not have produced this much compost. The black dashed line denotes the component level boundary of the farm. Inflows and outflows outside the boundary represent those at the farm level. The gray dashed lines denote relationships which were excluded from the study. Labels in italic signify factors not quantified, for which it was still worth identifying their place within the system. Figure 5.6 Nutrient flows (kg/farm/yr) that influences the nutrient balance of an enset- based system. The asterisk (*) after organic matter (IN2) denotes that this input likely came from an external source, as 0.43 TLU could not have produced this much compost. The black dashed line denotes the component level boundary of the farm. Inflows and outflows outside the boundary represent those at the farm level. The gray dashed lines denote relationships which were excluded from the study. Labels in italic signify factors not quantified, for which it was still worth identifying their place within the system. Figure 5.7 Nutrient flows (kg/farm/yr) that influences the nutrient balance of an enset- coffee system. The black dashed line denotes the component level boundary of the farm. Inflows and outflows outside the boundary represent those at the farm level. The gray dashed lines denote relationships which were excluded from the study. Labels in italic signify factors not quantified, for which it was still worth identifying their place within the system. Figure 5.8 Component level nutrient inflows, outflows and balances for N, P and K (kg/farm/yr) for an enset-cereal-vegetable representative farm. Figure 5.9 Nutrient flows (kg/farm/yr) that influences the nutrient balance of an enset- cereal-vegetable system. The black dashed line denotes the component level boundary of the farm. Inflows and outflows outside the boundary represent those at the farm level. The gray dashed lines denote relationships which were excluded from the study. Labels in italic signify factors not quantified, for which it was still worth identifying their place within the system. Figure 5.10 Component level nutrient inflows, outflows and balances for N, P and K (kg/farm/yr) for an enset-livestock representative farm. Figure 5.11 Nutrient flows (kg/farm/yr) that influences the nutrient balance of an enset- livestock system. Figure 5.12 Component level nutrient inflows, outflows and balances for N, P and K (kg/farm/yr) for a khat-based representative farm.

pg. 15

Figure 5.13 Nutrient flows (kg/farm/yr) that influences the nutrient balance of a khat-based system. The black dashed line denotes the component level boundary of the farm. Inflows and outflows outside the boundary represent those at the farm level. The gray dashed lines denote relationships which were excluded from the study. Labels in italic signify factors not quantified, for which it was still worth identifying their place within the system. Figure 5.14 Farm level nutrient inflows, outflows and balances for N, P and K (kg/farm/yr) across representative farms.

Equations

Equation 4.1 Land use percentage for each component. Equation 4.2 Average component percentages to equal 100. Equation 4.3 Macronutrient amount from output. Equation 4.4 Total sum of nutrient (TSN) for each component. Equation 4.5 Harvest index. Equation 4.6 Mean macronutrient amount from enset output. Equation 4.7 Macronutrient balance.

pg. 16

ACKNOWLEDGEMENTS

This research could not have been realized without the help of several inspiring people. I would like to express my sincere gratitude towards Dr. Gerrie van de Ven and Dr. Katrien Descheemaeker for their interest, guidance and welcoming to Plant Production Systems (PPS). In Ethiopia, thank you to my daily supervisor at Hawassa University, Drs. Beyene Mellisse. This thesis would not have been possible without your advice and feedback, thank you for taking an earth scientist under your agricultural wing. To carry out this research, thank you to the farmers across the Wondo Genet, Malga, Bule and Dale woredas, for welcoming this “faranji” (foreigner) into your farm and sharing your vast knowledge.

A special thanks to my supervisor at the University of Amsterdam, Dr. Erik Cammeraat. Thank you for trusting me in the realization of this project. I would also like to extend my gratitude to Dr. Boris Jansen, who will act as my co-assessor and second reader at the University of Amsterdam. Moreover, I want to thank my fellow earth scientists for the detailed feedback on my proposal and research workshop presentations. Constructive feedback is invaluable to the thesis process. At Wageningen University, I owe gratitude to fellow PPS students for challenging my proposal and inspiring me to persevere.

Finally, a special mention to Jolanda, Willem and Nina. We may be spread across the globe but with your combined support, my dreams feel infinitely within reach.

Nadine Joanne Galle Amsterdam

pg. 17

1. INTRODUCTION

Africa’s smallholders dominate the agricultural sector, which remains at the basis of developing economies. In Ethiopia, agriculture accounts for nearly 46% of gross domestic product (GDP), 73% of employment and almost 80% of foreign export earnings (Ethiopia Agricultural Transformation Agency [EATA], 2014). Nationwide food security and Ethiopian livelihoods are profoundly reliant on the success of this sector. In the Sidama and Gedeo zones of southern Ethiopia, the enset (Ensete ventricosum) and coffee (Coffea arabica) based home gardens have sustained millions of livelihoods for centuries. Enset is a species of the banana family where the pseudostem (not the fruit) is consumed. Its edible products kocho and bula are the region’s staple food and its leaves offer construction material and protein-rich livestock feed. Enset cultivation requires relatively low external inputs, has a large food per unit area capacity and holds a distinct resistance to drought. Coffee has long reigned as the dominant cash crop in these parts. In 2005, according to the Central Statistical Agency of Ethiopia (CSA, 2007), total coffee supplied to market from Sidama and Gedeo was 63,562 tons, which accounted for 63% of the regional (Southern Nations, Nationalities and Peoples’ Region [SNNPR]) and 23% of the national coffee production.

Owing to the two dominant perennial crops, these traditional home gardens are often referred to as ‘enset-coffee’ home gardens. These systems are characterized by the farming of annual and perennial agricultural crops and livestock in close association with trees and/or shrubs (Abebe, 2005; Kippie, 2002). Ninety percent of Ethiopian smallholders practice the home garden system, typically cultivating less than one hectare. Despite their small size, home gardens support dense populations, ensure consistent availability of multiple products and generate employment and income through multi-species integration (Kumar & Nair, 2004). Home gardens have long been established as stable farming systems managed by family labour with low external inputs. Efficient nutrient cycling within farms, offered by multi-species composition, conservation of bio-cultural diversity and product diversification, are some of the key factors contributing to the stability of these systems in SNNPR, one of the most densely populated regions in Ethiopia (Abebe, 2005).

In recent decades, the trend of home garden change driven by increasing population pressure induced land fragmentation, has led to rapid replacement of enset and coffee with khat (Catha edulis) (Mellisse et al., in prep.). Khat, a lucrative cash crop grown for its financial gain and chewed for its stimulating effects, has expanded at the expense of land allocated to enset, coffee and in some cases other food crops (e.g. cereals) and cash crops (e.g. vegetables) (Abebe et al., 2010). Compared to coffee, khat generates higher financial returns, uses less water and can be harvested multiple times a year. Mellisse et al. (in prep.) reported that the combined area share of enset and coffee covered more than 45% of the farms in four study districts (Wondo Genet, Melga, Bule and Dale) of Sidama and Gedeo zones in 1991. Two decades later it fell below 25% in both Wondo Genet and Melga, while the share of khat rose from 7% and 5% in 1991 to 36% and 33% in 2013, respectively (Mellisse et al., in prep.). Dale increased khat share by 0.9% in 1991 to 9% in 2013. In contrast, Bule experienced an expansion rate so low it

pg. 18

hardly warrants mention. Nonetheless it is evident the introduction of khat in the study area’s home gardens produced rapid change in its cropping and land use patterns.

The transition has called for the definition of five distinct home garden types: four enset- oriented (enset-based, enset-coffee, enset-cereal-vegetable, and enset-livestock) and one khat- based. Enset-oriented farms rely heavily on internal inputs of organic matter in the form of compost. Farms cultivating cereals, vegetables and khat are more dependent on external inputs of mineral fertilizers. In addition to releasing nutrients, compost improves soil structure and its water-holding capacity. These well-established internal nutrient flows are what sustain these home gardens; shifting to external input-only crops could alter these flows, induce deficiencies of nutrients lacking in mineral fertilizers and implicate the long-term stability of these systems.

It is argued the future of Sidama and Gedeo agriculture hinges on the expansion of khat monoculture (Mellisse et al., in prep.). Intensification without adequate restoration of soil nutrient supply may threaten this transition’s sustainability. Of the chemical processes involved in soil degradation, nutrient depletion is one of the most important as nutrient stocks are central to crop production (Syers, 1997). Nutrient ‘hotspots’ can reveal depletion, or may indicate an excess of unused nutrients which could be better utilized in other areas of the farm. The once ubiquitous enset-coffee home garden has transitioned into five distinct home garden types. The divide within a confined study area offers a unique opportunity to compare systems under the same climatic and biological conditions. As such, analyzing the implications of recent transition in home garden systems could highlight potential nutrient-related consequences of the introduction of khat. Figure 1.1 displays a traditional home garden, enset plants, harvest of the enset plant and fresh khat leaves.

1.1 RESEARCH WITHIN THE CASCAPE PROJECT

Despite Ethiopia’s technological advancements and accelerated agricultural growth in recent years, low agricultural productivity persists. The Government of Ethiopia adopted the five year Growth and Transformation Plan (GTP) in its hopes to eradicate poverty. Within GTP, the Agricultural Growth Programme (AGP) was established to realize full food security and support high economic and export growth. Scaling up best practices has the highest priority. The ‘Capacity building for scaling up of evidence-based best practices in agricultural production in Ethiopia’ (CASCAPE) project was designed to support the Ethiopian government in increasing agricultural productivity for smallholder farmers by identifying and disseminating best practices. Funded by the Ministry of Foreign Affairs of The Netherlands through the Dutch embassy in Addis Ababa, CASCAPE collaborates with six Ethiopian universities (Addis Ababa, Bahir Dar, Haramaya, Hawassa, Jimma and Mekelle) and ALTERRA at Wageningen University and Research Centre (WUR). Working closely with regional research institutes and Bureaus of Agriculture, CASCAPE aims to strengthen shareholder linkages and improve sustainable farming strategies.

pg. 19

Figure 1.1 From left to right: 1) the traditional home garden with grazing land in the foreground and behind that, the homestead, then enset infields and coffee/annual cereals and vegetable/khat outfields (Galle, 2015), 2) Enset plants, with Drs. Beyene Mellisse and our translator for scale (Galle, 2015), 3) Women harvesting the enset plant (Mellisse, 2015), 4) Bunches (zurba) of fresh khat leaves (CCTV Africa, 2014).

1.2 SOCIETAL AND SCIENTIFIC SIGNIFICANCE

To realize an ecologically sustainable and favourable socio-economic future for the people of Sidama and Gedeo, their home gardens must be resilient to this change in cultivation. Currently, little is known on the nutrient accumulation, losses and management of these systems. Research on the topic is either out of date (Eyasu, 1997), on a continental scale (Stoorvogel, Smaling & Janssen, 1993) or at the national level (Roy et al., 2003). Moreover, Ethiopia has 12 diverse agro-ecological zones, rendering much of the existing research spatially

pg. 20

irrelevant and unsuitable for comparative analysis (Abera, 2013; Abrham, 2014; Haileslassie et al., 2006; Kiros et al., 2014).

Furthermore, when confronted with new crop cultivation (e.g. khat, annual cereals and vegetables), farmers turn to nation-wide blanket recommendations regardless of local soil conditions. Reorienting extension efforts of blanket prescriptions by presenting home garden type-specific information can empower smallholders to diagnose nutrient accumulations and soil degradation. A knowledge gap exists to better understand the altered inflows and outflows that impact the nutrient balance of these evolving systems, ideally with site-specific expertise.

Traditional home gardens are highly dependent on organic fertilizers in the form of compost for enset, coffee and some annual cereals and vegetable fields. But quantification of nutrient amounts are lacking. According to Mwangi (1996), inorganic fertilizer use is assumed to be relatively low. Wallace & Knausenberger (1997) have even argued for increased inorganic fertilizer use with minimal environmental consequences, but socio-economic factors, lack of credit and pricing policy hinder farmer accessibility. Investigating farm-specific inorganic fertilizer use and its interactions with other internal and external homegarden inputs will support the accurate quantification of the nutrient balance of these systems before they transition entirely.

Sidama and Gedeo are at a turning point. The diversity of systems veering away from traditional enset-coffee home gardens is novel and vastly under researched. As khat monoculture is set to increase in coming decades, the distinct home garden types have never been more divided. The opportunity for comparative analysis at present is exemplary. And with this transition already underway for over two decades (Mellisse et al., in prep.), the demand for this research has never been greater.

1.3 OUTLINE OF THE THESIS

The research is structured as follows. Chapter 2 presents the research aims and questions. Chapter 3 consists of a brief overview of the relevant theories and existing research on nutrient balance assessments. Chapter 4 gives an overview of the study area, data collection and the methodologies used in this research. Chapter 5 shows the results of the quantification and comparisons. In Chapter 6 the results and methodology are discussed in relation to previous research. In addition, methodological improvements, suggestions for further research and management recommendations are provided. Chapter 7 completes the thesis with the conclusions. The appendix features additional figures and tables that are used in this research.

pg. 21

2. RESEARCH OBJECTIVES AND QUESTIONS

The objectives of this research are:

1. Produce representative farms for each home garden type based on farm component (e.g. crops, livestock) prevalence.

2. Quantify N, P and K inflows, outflows and internal flows for the representative farms across study districts of Wondo Genet, Malga, Dale and Bule for the cropping season 2014/15.

3. Compare representative farms based on component level and farm level nutrient balance assessments to assess nutrient depletion or accumulation under current nutrient management.

4. Improve and broaden understanding of inflows, outflows and internal flows that influence the nutrient balance of transitioning home garden systems and recommended future management actions.

The associated research questions are:

1.1 Based on the representative farms, what farm components are most significant in each home garden type?

2.1 How do N, P and K inflows and outflows differ amongst the home garden types?

3.1 How do nutrient balances compare across home garden types at component level, and at farm level?

3.2 Where do “hotspots” of nutrient depletion and/or accumulation exist?

4.1 What future management actions can be taken to improve nutrient management?

pg. 22

3. THEORETICAL FRAMEWORK

Overview of the existing literature provides the basis for this study’s theoretical framework. In this chapter, key terms are defined (3.1) and the inflows (3.2), outflows (3.3) and internal flows (3.4) of the home garden system are described.

3.1 DEFINITION OF CONCEPTUAL TERMS

Conceptual terms necessary to answer research questions and conduct methods are explained in this section.

3.1.1 NUTRIENT BALANCES

Much like a financial balance, nutrient balances reveal surpluses or deficits. Dobermann (2005) expresses a surplus or deficit as either a measure of net depletion (output > input) or enrichment (output < input). A surplus, or an accumulation of high levels of nutrients, is often attributed to negative environmental consequences, in which case, the nutrients are considered pollutants. A moderate surplus, however, could result in improved soil fertility. - Nutrients can be exported from farms in the form of runoff (P and some N), leaching (NO3 and - + some P) or its gaseous form via denitrification (NO3 to N2) or volatilization (NH4 to NH3). A deficit, or nutrient loss, can indicate land degradation and lead to gradual soil depletion. Ultimately, both outcomes can render agricultural practice unsustainable in long-term.

By examining inputs, outputs and storage processes of farming systems, nutrient balances can help in managing nutrients by identifying production goals and opportunities for improvement (Gourley et al., 2007). Balancing nutrient inputs and outputs can reduce undesirable off-farm nutrient consequences (e.g. eutrophication caused by excessive nitrogen runoff) and reduce expenditure on farm inputs (e.g. fertilizers and feed supplements). The balances are produced for various spatial and temporal boundaries. In brief, a balance tracks inputs, outputs and stores of a defined system over a fixed period of time, such as a specific year. Balances can range from broad farm-gate analyses to those at specific field-level to detailed soil-level studies. The purpose of the balance determines the degree of data detail necessary. Naturally, this also works in the opposite fashion, where the extent of data availability limits the balance’s detail.

Each level of nutrient balance—farm-gate, field and soil—has its benefits and limitations. Farm- gate nutrient balances (FGB) are producible with readily available data, easily repeatable and simple to communicate (Öborn et al., 2003). FGBs also have the capacity to account for multiple nutrients, calculate outcomes financially and are useful for farm budgeting. However, the FGB can overlook depletion caused by flows of nutrients within the farm (Cherry et al., 2008). Internal flows can be significant; with shortages in some areas and accumulation in others. Fluctuations in local conditions (e.g. climate and soil fertility) and nutrient fluxes (e.g. biological nitrogen fixation (BNF), atmospheric deposition (DEP) and leaching) are typically not accounted

pg. 23

for because such detail is too much for the purposes of a FGB. Internal flows are better analyzed in a field-level balance, which examines the balance at the soil-surface level on each field within a farm. Field-level balances consider DEP, BNF, leaching and nutrient content of manures and crops, but typically rely on estimates and assumptions for these components. Localized surpluses and deficits can be better identified and managed using a field-level balance (Cherry et al., 2008). Soil-level balances measure denitrification, volatilization and lateral transport. It is the only balance to accurately account for spatial and temporal aspects of fluxes (Öborn et al., 2003), but requires high quality data. Soil-level balances are a useful tool for site- specific research and development. Only when it is representative of the greater system, can soil-level balances identify processes where problems occur and follow the fate of nutrient sources (Wander, 2015).

3.1.2 COMPONENT LEVEL NUTRIENT BALANCE

First, a balance must delineate strict temporal boundaries and as such our balance was annual, examining the 2014/15 cropping season. Second, partial nutrient balances were used. Agricultural fields tend to have residual nutrients and because of the difficulty in measuring all individual output pathways into the environment, residuals were assumed to be zero (Jackson- Smith, 2010). Third, a balance must adhere to spatial boundaries; therefore a component level nutrient balance was used. The component level approach lies between the coarser FGB and a more comprehensive field-level balance. The fields are grouped together into components by land use. For example, the annual cereals and vegetables component groups together maize, barley, onion and cabbage fields. For diversified farms cultivating multiple crops in a small area, like a home garden, simply using a FGB would underestimate the influence of internal flows. Field-level assessments can identify movement of nutrients within farms, but require a detail not feasible in the time span of this project. Grouping fields together in components is practical while it still provides a general indication of environmental performance and detailed insight of internal nutrient flows. The choice behind a component level balance is especially relevant for home gardens in the study site, where land use allocation has changed and distinct home garden types have emerged, thus disrupting internal flows and calling for their comparison. The farm components in this study are enset, coffee, enset + coffee intercropping, annual cereals and vegetables (ACV) and khat. Livestock was also a component.

3.1.3 NUTRIENT FLOWS

Mellisse et al. (in prep.) produced a schematic model of nutrient outflows and inflows of five different home garden types (Figure 3.1). In the model, nutrient outflows from the farm include market, which are sales of livestock produce (e.g. meat, milk) and exported crops (chiefly, coffee, vegetables and khat). Several nutrient losses are also defined, including: losses from front grazing land by leaching, volatilization and erosion (L1), losses from livestock (L2), losses from the manure heap by leaching and volatilization (L3), losses during application of manure to fields (L4), losses from the home garden fields by leaching and volatilization (L5).

pg. 24

Nutrient inflows to the farm include DEP, BNF, purchased food crops, livestock and farm inputs (Market), and cattle which are taken from other farms for fattening purposes (Fat) (e.g. feeding the cattle for three months and then returning them to the owner) (Mellisse et al., in prep.).

Figure 3.1 Schematic model of nutrient inputs and outputs across the five home garden types, including inputs: atmospheric deposition (DEP), biological nitrogen fixation (BNF), purchased food crops, livestock and farm inputs (Market), cattle which are taken from other farms for fattening purposes (Fat) (e.g. feeding the cattle for three months and then returning them to the owner). The model also shows outputs: losses from front grazing land by leaching, volatilization and erosion (L1), losses from livestock (L2), losses from the manure heap by leaching and volatilization (L3), losses during application of manure to fields (L4), losses from the home garden fields by leaching and volatilization (L5). Nutrient flows on individual home garden include: feedstuffs taken from front grazing yard (f1), cow dung left in front grazing yard (f2) (especially during day time since animals are tied up in grazing yard), milk and meat consumed by the family (f3), collection of farm yard manure (FYM) (f4), application of FYM in different land use type (f5), feedstuff taken from different land use type (especially, enset leaves) by livestock (f6), household waste added to manure heap (f7), family consumption of both perennial and annual crops (f8) (Mellisse et al., in prep.).

The schematic model (Figure 3.1) also displays internal nutrient flows within an individual home garden. It is important to distinguish between internal and external inputs. Internal inputs are on-farm resources such as manure, enset leaves and grasses. External inputs are off-farm

pg. 25

resources purchased for use on the farm such as livestock fodder, chemical fertilizers, insecticides and/or pesticides. If manure or livestock fodder was purchased, rather than produced on the farm where it is applied, it is considered an external input. In this study, nutrient flows and nutrient balances allow understanding the interactions between home garden components.

3.2 INFLOWS IN TO THE HOME GARDEN SYSTEM

In southern Ethiopia, farmers’ use of agricultural inputs is highly dependent on their accessibility. Ethiopia’s Bureau of Agriculture, the open market, NGOs and neighbouring farms all supply farmers with inputs (Dessalegne et al., 2012). This section highlights only the major nutrient resource inflows that are considered for the farm level nutrient balance assessment. Atmospheric deposition (DEP) and biological nitrogen fixation (BNF) are excluded as inflows. DEP can occur in two forms: wet deposition (rain and fog) and dry deposition (gases and particles, without aid of precipitation). BNF is dependent on several soil factors. For example, the presence of phosphorus, the presence of appropriate Rhizobia and pH. The application of fertilizers is by far the most common way to supply crops with nitrogen. However, possibly more sustainable practices such as crop rotation with symbiotic N-fixation by leguminous crops or planting them alongside N-fixing crops are being used as well. For the partial nutrient balance assessment in this study, DEP is excluded due to difficulties in its accurate estimation (Munters, 1997). BNF is excluded, as only few legumes are grown in the study areas. Codes assigned to each inflow are not intended to be in numerical order.

3.2.1 MINERAL FERTILIZER (IN1)

- Nitrogen is one of the most abundant elements on earth, but only in the form of nitrate (NO3 ) + and ammonium (NH4 ) is it available for plant uptake. The application of mineral fertilizers is by far the most common way to supply crops with nitrogen. Dessalegne et al. (2012) report low mineral fertilizer use in Ethiopian home gardens. Despite projects like the Development of Competitive Markets (DCM), reforms designed to encourage private sector participation in fertilizer distribution, fertilizer use has remained low. The reasons are its high cost, unavailability, limited knowledge about its benefits and little information on how to properly apply it (Wallace & Knausenberger, 1997). Devaluations of domestic currency and lack of credit can also constrain fertilizer use in already impoverished areas. This is especially true in unirrigated, rain-fed agricultural zones, which are considered to have high risk.

However, upon completion in 1995, DCM did increase fertilizer sales drastically: from 132,000 tonnes in 1993 to 236,000 tonnes in 1995 (Wallace & Knausenberger, 1997). Kiros et al. (2012) reported Ethiopian fertilizer use rose to 7 kg/ha in 1997. This is comparable to the Sub-Saharan African average of 6 kg/ha but still very low compared to the global average of 78 kg/ha at that time (Makken, 1993). Results from DCM are nearly two decades old and with a lack of more recent data it is difficult to deduce current trends in inorganic fertilizer use. However, the price of inorganic fertilizer is still high due to its foreign production and poorly developed

pg. 26

infrastructure in Ethiopia. Abate et al. (2015) indicated that inorganic fertilizer use remains on the rise. Their research showed nationwide consumption of N and P for fertilizing maize at 20,000 tonnes in 2004 to 68,000 tonnes in 2013—a more than 3-fold increase (Abate et al., 2015). N and P accounted for roughly 67% and 33% of this growth, respectively.

All of Ethiopia’s mineral fertilizer is imported (Abate et al., 2015). Currently, the most commonly used nitrogen fertilizer in Ethiopia is urea. It can be inexpensively manufactured and is widely applicable to nearly all crops. Urea holds 46% nitrogen content. However, urea is highly soluble in water and measures to limit nitrogen runoff should be prudently undertaken. The nation’s most commonly used phosphate fertilizers are triple superphosphate (TSP) and diammonium phosphate (DAP). Both fertilizers are in dry form and when dissolved have pH values of 1.5 and 8.0, respectively. On basic soils TSP might be more effective, while on acidic soils DAP is hazardous as direct contact with seeds may cause seedling damage. TSP may be less favourable economically as it is more costly to produce. In Sidama and Gedeo agriculture, urea and DAP are available while TSP is not.

While high external input or even industrial agricultural nations are rightfully concerned about the negative environmental consequences of excessive fertilizer use, Kelly and Naseem (2004) argue Ethiopia faces negative environmental impacts of too little fertilizer use. Although environmental damage from too little fertilizer is unlikely, its on-farm effects have serious implications. For example, a soil that receives little to no inputs can rapidly lose nutrients, a process known as nutrient mining. This is especially true if inadequate biomass production limits nutrient recycling for future plantings.

3.2.2 EXTERNAL LIVESTOCK FODDER (IN4)

External fodder (IN4) comprises of sugarcane tops and wheat bran. These fodder sources often supplement internal fodder. Ethiopia strives to be one of the world’s top 10 sugar producers by 2023 and sugarcane tops are abundant in sugar-growing countries. Through the state-run Ethiopia Sugar Corporation (ESC), the Government of Ethiopia (GOE) has invested in new processing factories, revitalizing older factories and expanding sugar cultivated land to boost sugar production (Francom, 2015). One hectare of sugarcane can yield 30 tons (fresh weight) of tops (Mahala et al., 2013). Sugarcane tops are primarily fed to fatten livestock, rather than provide nutrients. They are highly palatable and can often sustain cattle with little protein supplement (Leng & Preston, 1976).

To livestock, wheat bran is also very palatable. As a by-product of the milling industry, wheat bran is diverse. Mixed wheat bran is widely considered of better quality due to its good proportion of flour and husks (Gebremedhin et al., 2009). Coarse bran has poor nutritional value while fine bran could cause bloating in livestock. Farmers near Hawassa prefer fine bran for fattening animals and coarse bran for dairy cattle (Gebremedhin et al., 2009).

pg. 27

3.3 OUTFLOWS FROM THE HOME GARDEN

This section highlights only the major nutrient resource outflows relevant to the local home garden systems. Outflows from fields such as leaching, gaseous loss and erosion were not expounded for this assessment due to lack of data. The following outflows were taken into account for the farm level nutrient balance assessment. These outflows are exported crops sold off-farm, whole livestock leaving the farm and livestock product sales. Codes assigned to each outflow are not intended to be in numerical order.

3.3.1 REMOVAL IN HARVESTED PRODUCTS SOLD OFF-FARM (OUT5)

Several different perennial and annual agricultural crops are grown in the Sidama and Gedeo zones. The traditional home gardens are dominated by enset and coffee. Accompanying these two main crops are vegetables (e.g. onion, cabbage) and some annual cereal crops (e.g. maize, barley). Khat has been increasingly cultivated for its economic lure in recent years, while at the expense of enset and coffee. To quantify and compare inputs and outputs at the farm level, the crops most often sold are cash crops: onions, cabbage, coffee and khat. On some occasions, kocho or enset leaves will be sold off-farm. Fibrous enset leaves are multi-functional and could be sold as building material or textile for clothing.

3.3.2 LIVESTOCK OUTPUT (OUT3)

Exporting whole animals is another potential resource flow out of the home garden system. Most commonly, chickens, goats, sheep and cattle are sold for consumption. However, whole animals for utilization can also be sold and/or traded amongst farmers. Livestock are primarily sustained by enset leaves and grasses (Mellisse et al., prep.). On occasion, their diet is supplemented with purchased external fodder (e.g. sugarcane top, wheat bran). These nutrient inputs cycle back to crop fields in the form of manure mixed in compost (IN2).

From the milk produced by the cattle, part is sold and part accounts for nutrient loss from the farm. Butter is also made from the milk, but has been grouped together with milk for this assessment. Eggs are also sold and result in nutrient losses. In the study area, meat is not sold separately, only as whole animals leaving the farm.

3.4 INTERNAL FLOWS IN THE HOME GARDEN SYSTEM

This section describes the internal nutrient flows relevant to the home garden system. Codes assigned to each internal flow are not intended to be in numerical order.

3.4.1 ORGANIC MATTER (IN2)

In Ethiopia, organic fertilizers can be categorized into animal manures and compost. In Sidama and Gedeo, livestock manure is argued to be the principal farm input to crops (Mellisse et al., in

pg. 28

prep.). Applying manure improves soil fertility and its physical condition (Elias, 2002). With the aim to supply nutrients, manure is regularly applied. However, its effect varies with application amounts and manure quality, and is often dependent on livestock and labour availability, which is necessary to transport manure onto fields (Kiros et al., 2012). Use of manure on crops competes with non-farm uses. For example, increasing shortage of fuel wood forces rural Ethiopians to burn dried cattle dung. Kiros et al. (2012) found this to deprive soil of an important source of organic matter and nutrients.

Compost is the decomposed organic waste produced from crop residues, animal manure, household waste and sludge. It is stabilized by macro- and micro-organisms through aerobic, semi-aerobic and anaerobic biological processes. In Sidama and Gedeo, composts made mainly of cattle dung and household refuse is most commonly used. Utilizing human excreta is still widely considered taboo. Compost in itself is not a rich source of nutrients, but acts as an important soil amendment by increasing microbial activity and soil fertility. Like manure, application amounts vary according to labour availability. As a result, fields located close to the homestead generally receive more compost compared to fields further away. Compost is typically collected, decomposed and stored in an outdoor pile close to the homestead. From there it is distributed to fields. Dessalegne et al. (2012) stipulated further research is essential to survey if compost alone is enough to increase home garden productivity.

3.4.2 INTERNAL LIVESTOCK FODDER (IN3)

To feed livestock, internal fodder (IN3) consists of enset leaves and grasses collected from enset, coffee and khat fields. Typically, internal fodder remains within the farm by means of livestock manure that is applied on the crop fields, which is a characteristic of a closed-loop system. Together with manures, crop residues can replenish the essential macronutrients; contribute to maintaining soil organic matter and the soil’s structure. Exporting these resources off-farm can have negative nutrient-related consequences, should they not be replaced with inorganic fertilizer means. For example, in Ethiopia, complete removal of all crop residues (internal fodder) is estimated to remove 101 and 168 kg/ha/yr of N and P nutrients, respectively (Kiros et al., 2012). While fattening supplements via external fodder are crucial in livestock feed, the role of internal fodder should not be undervalued.

3.4.3 REMOVAL IN ALL HARVESTED PRODUCTS (OUT1)

Nutrient removal in all harvested products considers all crops that are produced, not only those that are sold off-farm. Crops most likely to remain within the home garden are enset products (e.g. kocho, bula) and cereals (e.g. barley, maize). These crops are presumably consumed by the household.

pg. 29

3.4.4 REMOVAL IN CROP RESIDUES (OUT2)

Crop residues are relevant only on maize and barley fields where stover and straw remains on the soil after harvest. The harvest index (HI) was used to determine these above-ground crop residues. On enset, coffee and khat fields, all grasses are collected for internal livestock fodder (IN3) and no crop residues remain. Regarding vegetables, onion and cabbage are both uprooted. As such, these fields can also be considered a complete harvest.

3.4.5 HOUSEHOLD LIVESTOCK CONSUMPTION (OUT4)

Household crop consumption was not explicitly asked for in the input/output survey, but household livestock consumption was. However, the survey only asked about the household consuming whole animals instead of livestock products. Since household consumption of milk and eggs was not explicitly requested, it has been excluded from the nutrient balance assessment.

pg. 30

4. METHODOLOGY

To analyze nutrient management and flows on transitioning home gardens, component level and farm level macronutrient balances are used to compare the nutrient amounts entering, leaving and circulating (within) the farm. The methodology considers the system over the 2014/15 cropping season. Nutrient balances can be indicators of (i) a nutrient surplus (inputs > outputs), leading to an accumulation, (ii) a deficit (outputs < inputs), depleting nutrient reserves and heightened risk of reduced crop yields due to nutrient mining – the unreplenished nutrient removal by crops, or (iii) a neutral balance (Cuttle, 2002). The research strategy to quantify nutrient inflows, outflows and internal flows and to calculate the farm level and component level nutrient balances are described here.

4.1 STUDY AREA

Ethiopia has a complex history of dividing its country. As of 2015, Ethiopia has 9 regional states (kililoch) and two chartered cities (Addis Ababa, Dire Dawa). Kililochs are based on ethnic territoriality and further subdivided into 68 zones. Some zones are further divided into districts (woreda), which are then split into municipalities (kebele). Kebele are the smallest unit of administrative division in Ethiopia.

The research was conducted in the Sidama and Gedeo zones of the Southern Nations, Nationalities and Peoples’ Region (SNNPR) kililoch (regional state) in southern Ethiopia (Figure 4.1). Encompassing 7,672 square kilometers (Abebe, 2005), Sidama is located at 5°45’-6°45’N latitude and 38°-39°E longitude and home to 3.5 million inhabitants (CSA, 2007). The area is densely populated with over 450 people per km2 (CSA, 2007). Some 95% of the inhabitants speak Sidaamu Afoo, the district’s primary first language. In contrast to the majority of northern Ethiopia’s arid landscape, Sidama is largely lush and green with rolling hills and fertile valleys. Sidama is subdivided into 19 woredas (district), of which three (Wondo Genet, Malga, Dale) were studied. The woredas were chosen based on population density, agro-ecology and access to markets.

Sidama surrounds the city of Hawassa. At an altitude of 1665 m a.s.l., Hawassa serves as SNNPR’s capital. Hawassa has 165,275 inhabitants representing over 50 ethnicities. Nearly half of the population resides in Hawassa’s nearby kebeles (neighbourhood). The city lies adjacent to Lake Awasa, the smallest of the Great African Rift Valley lakes. Its fish combined with that of the neighboring Abaya Lake in Gedeo provide a robust local fishing industry.

Gedeo is subdivided into eight woredas, of which one (Bule) was studied. Again, Bule was selected for its population density, distinct agro-ecology and distance to markets (add table). Its isolation contrasts well with Sidama’s selected woredas, rendering a diverse and representative study area. Gedeo is 1,347 square kilometers (Kippie, 2002), has a population of 0.84 million and is located at 5°-7° N latitude and 38°-40° E longitude (CSA, 2007). Gedeo shows similar variation in elevation to Sidama, with a range of 1268 (at Lake Abaya) to 2993 m.a.s.l. (at Haro

pg. 31

Wolabu Pond). Gedeo zone is named after its Gedeo people, which predominantly speak the Gedeo language. Although the languages are alike, the Gedeo people have a distinct culture compared to the Sidama people. However, both zones share identical agricultural economies based on cultivating enset and coffee within traditional home gardens.

The zones also share comparable bimodal rainfall distribution, ranging from 1200 to 2000 mm per annum (Abebe, 2005). The long (June to September) and short (March to May) rainy season create favourable conditions for the dominant perennial-based home garden systems. Sidama and Gedeo cover two agro-ecological zones (Table 4.1). Both zones support different agriculture and lifestyles.

Table 4.1: Agro-ecological zones with characterizing altitude, rainfall, temperature and predominant perennial crops (Mellisse et al., in prep.)

Altitude Average annual Average annual Agro-ecological zone (m.a.s.l.) rainfall (mm) temperature (oC)

Moist mid-altitude, subtropical zone 1500-2300 1200-1600 16-22 (Amharic: woinadega) Moist highland, cool zone 2300-3200 1600-2000 15-19 (Amharic: dega)

In SNNPR, there are 116 meteorological stations recording climate data. Stations in Hawassa and Arba Minch (270 km south of Hawassa) are synoptic (large-scale) receiving satellite data and recording all weather elements. The other 114 stations vary and are less detailed. Altitude and rainfall are the main determinants of climate in the region. Dominant soil types are Nitosol, Cambisol and Lithosol (Tsegaye, 2001).

pg. 32

Figure 4.1: Location of study districts (woredas: Wondo-Genet, Malga, Dale, Bule) within Sidama and Gedeo zones of Southern Nations, Nationalities and Peoples’ Region (SNNPR). The legend displays Ethiopia’s nine regional states and two charted cities.

pg. 33

4.2 FARM TYPOLOGIES

Home garden systems are diverse. To better analyze their differences; Mellisse et al. (in prep.) constructed a farm typology. Different farms with similar characteristics were categorized into five types—four enset-oriented (enset-based, enset-coffee, enset-cereal-vegetable, enset- livestock) and one khat-based. To construct the typology, data from the 240 surveyed households and multivariate statistical analysis (MSA) was used. The farm types were identified based on area shares of enset, coffee, khat, annual cereal and vegetables and grazing land.

4.3 DATA COLLECTION

The data for a partial macronutrient balances and MNUE was acquired through household surveys. In 2012/13, 240 households across the four study woredas were surveyed. The heads of the households were asked to report the year of khat introduction, land allocation to various perennial and annual crops, total land holdings and livestock herd size (Mellisse et al., in prep.). Demographic characteristics such as family size and level of education, production objectives, sources of income, constraints to crop production and livestock rearing and dependency on the market for family food, were also requested. Secondary data at the woreda and kebele level were collected on population density, population size and kocho, coffee and khat prices. From this analysis, the farm typologies were created.

Of these 240 households, a sub-sample of 63 households was selected for the detailed input/output survey. In this survey, data on macronutrient inputs, outputs and stores of the 2014/15 cropping season for farmer’s home gardens was recalled. Household surveys also required farmers to specify livestock kept, died, consumed or sold. Livestock type and gender was also asked. The translated English version of this household survey questionnaire is available in Appendix 7.3. The data collected from these 63 surveys was entered into Excel for further analysis. This research project compiled all macronutrient inputs and outputs at the component and farm level, conceived representative farms and conducted partial macronutrient balances for each home garden type.

In addition to detailed survey data, composite samples of crops, internal livestock fodder and home garden compost were taken for nutrient content analysis prior to my arrival in Ethiopia. The samples were analyzed in the laboratory facilities of Hawassa University College of Agriculture and the Wondo Genet College of Forestry and Natural Resources. Based on nutrient values and dry matter, the amount of nutrients transported from both component and farm level was quantified. A complete table of the nutrient content of all output is available in Appendix 7.2. The nutrient content for cabbage, milk and eggs were taken from literature (The National Agricultural Library, 2015; Myburgh et al., 2012; Roe et al., 2012). The nutrient content of animals leaving the farm was based on Van Heerden et al. (2002) and the Agricultural Research Council [ARC] (1984). For external livestock fodder, which includes sugarcane top and wheat bran, the nutrient content was also taken from literature (Heuzé et al., 2015a; Heuzé et al., 2015b).

pg. 34

Literature on typical sub-Saharan African macronutrient inputs and nutrient balances for similar smallholder subsistence systems was studied to develop a frame of reference for detecting outliers. Outliers may indicate either variability in the measurements or experimental error. Selection criteria for identifying outliers were based on Ethiopia’s blanket recommendation of 100 kg/ha of DAP and urea fertilizers. As Ethiopian authorities report fertilizer recommendations in kilograms, these data sets were converted into actual elemental nitrogen and phosphorus nutrients for standardized comparisons. Thus, Ethiopia recommends 63 kg/ha of N (DAP = 18 kg of N, urea = 45 kg of N) as DAP is 18% N and urea is 45% N. To account for potential over-fertilization, any application of more than 125 kg/ha of N was isolated as an outlier, but no outliers were revealed in the dataset.

4.4 EXPERIMENTAL DESIGN

In this section the experimental design and analytical techniques are described. The representative farm for each home garden type is explained, as well as the approach for extracting the nutrient amount from inputs and outputs and assessment of the macronutrient balance for both crop and livestock. As Ethiopian farmers tend to use their own units, a conversion table for all local units to kilograms can be found in Appendix 7.1.

4.4.1 THE REPRESENTATIVE FARM

Drawing from the farm typology designed by Mellisse et al. (in prep.) for each home garden type a representative farm was formulated. The first phase is to categorize crops and distinguish components of a home garden. The five determined components are: enset, coffee, annual cereals and vegetables (ACV, including maize, barley, onion and cabbage), khat and livestock. Annual cereals and vegetables were grouped based on their similarity in inputs and the fact that they are annual crops. Despite their opposing roles as food and cash crops, annual cereals and vegetables are often found together. A trend of increasing ACV area to meet household dietary and income needs was observed and resulted in a separate enset-cereal- vegetable home garden type (Mellisse et al., in prep.).

Second, a criterion to exclude or include the type of component from or to a specific representative farm was set based on its presence. Accordingly, a component represented in 50% or more of the surveyed farms was retained and excluded otherwise. For example, in the enset-coffee home gardens, 14 of the 18 farms cultivated coffee, so coffee was included. The third phase is to determine the proportion of land allocated to each component in its representative farm. For this, the allocated land area for each selected component over the total farmer-reported farm size was taken.

.//01.234 /.54 .63. 706 108905352 !"#$ &'( (ℎ"/ℎ") = (Equation 4.1) 202./ :; 7.68 <=>3

Note: FR denotes farmer-reported.

pg. 35

WHAT’S IN AND WHAT’S OUT In this study, all fruits and some vegetable crops were not taken into account. Crops such as bananas, avocados, mangoes, guava, potatoes and faba beans were excluded due to their negligible presence in the home gardens, as only 8 of 63 home gardens cultivated these crops.

Equation 4.1 is repeated for each component on each farm and averaged. This constitutes the average land use of each component within the representative farm. The averages will not add up to 100%, as excluded components are not accounted for. Therefore, the following formula (Equation 4.2) is applied to calculate the altered percentages.

x % + y % z % = must equal 100%

(100/N) ∗ N % = 1 (100/N) ∗ P % = " % (Equation 4.2) (100/N) ∗ Q % = R % " % + R % = 100%

Where x = component x y = component y z = original percentage sum, before alterations a = altered percentage for component x b = altered percentage for component y

The representative farm approach has the capability of providing insights into an otherwise complex farm typology.

To gain greater understanding of the livestock component, Tropical Livestock Unit (TLU, 250 kg bodyweight) was calculated for each representative farm to indicate the potential influence of livestock within different home garden types. TLU are livestock numbers converted to a common unit (Table 4.2).

Table 4.2: Tropical Livestock Unit (TLU) conversion chart (FAO, 1987).

Species TLU conversion factor Cattle 0.70 Sheep 0.10 Goat 0.10 Chicken 0.01 Horse 0.80

pg. 36

4.4.2 QUANTIFYING NUTRIENT FLOWS

Stoorvogel and Smaling (1990) pioneered the methodology behind nutrient balance assessments. Subsequent studies have modified the methods to fit study objectives and research location. Their original model included five input and five output processes: mineral fertilizer; organic matter, comprising manure and household refuse and leaf litter; atmospheric deposition (DEP); biological N-fixation (BNF); and sedimentation (inflows) and removal in harvested products; removal in crop residue; leaching; denitrification; and water erosion (outflows). Input and output processes quantified for this research varied on the component and farm level and were adapted to the local context.

On the component level, the major input flows quantified were 1. mineral fertilizer (IN1), 2. organic matter (IN2), comprising of manure and household refuse, 3. internal fodder (IN3), 4. external fodder (IN4).

The key output flows quantified were 1. removal in harvested products (OUT1), 2. removal in crop residue (OUT2), 3. whole livestock and livestock products sold off-farm (OUT3), 4. household livestock consumption (OUT4).

Input flows; DEP, BNF and sedimentation were excluded. DEP is excluded due to difficulties in its accurate estimation and lack of local data (Munters, 1997). BNF is excluded, as only few legumes are grown in Sidama and Gedeo. Sedimentation is not relevant as there are no irrigation schemes or flood plains in the study area (Elias et al., 1998). Output flows; leaching, denitrification and water erosion were excluded based on lack of regional-specific measurements which are also subject to temporal variability. Household consumption of livestock products (e.g. milk and eggs) was excluded due to lack of explicit data. Manure as a direct output from livestock was excluded because no composite samples were taken of fresh manure.

On the farm level, the major input flows quantified were 1. mineral fertilizer (IN1), 2. external fodder (IN4).

The key output flows quantified were 1. removal in harvested products sold off-farm (OUT5), 2. whole livestock and livestock products sold off-farm (OUT3).

pg. 37

4.4.3 MACRONUTRIENT INPUT

The survey data supplied DAP, urea and compost inputs in kilogram (kg) or gimbola (9.78 kg). The formulas used to extract macronutrient amount from these inputs are presented in the workflow below (Figure 4.2):

Figure 4.2: Conceptual workflow showing steps and formulae used to extract macronutrient content from input.

The third step in the workflow mentions the nutrient content of the inputs (Table 4.3). Livestock totals were derived using the same conceptual workflow (Figure 4.2) to calculate the nutrient content of internal and external fodder.

Table 4.3: Macronutrient content (mean ±SD) for the four input processes employed in calculating partial nutrient balances.

Input Code and %DM N (%) P (%) K (%) Reference (N) Reference (P/K) process nutrients DAP IN1 - 18 20 - (Mitchell, 2008) (Mitchell, 2008) (NH4)2HPO4 (N & P)

UREA IN1 - 45 - - (Mitchell, 2008) (Mitchell, 2008) CO(NH2)2 (N) (Wondo Genet Organic IN2 (HU Agricultural 26.37 0.83 0.03 0.29 College Soil Lab, matter (N, P & K) College, 2015) 2015) (HU Agricultural (Wondo Genet IN3 Grass 33 1.63 0.49 1.96 College Soil Lab, College Soil Lab, (N, P & K) 2015) 2015) (HU Agricultural (Wondo Genet Enset IN3 1.32 0.45 4.6 13.7 College Soil Lab, College Soil Lab, leaves (N, P & K) (±0.22) (±0.09) (±0.48) 2015) 2015) Sugarcane IN4 (Heuzé et al., (Heuzé et al., 26.8 0.78 0.12 1.87 tops (N, P & K) 2015a) 2015a)

Wheat IN4 (Heuzé et al., (Heuzé et al., 87 2.77 1.11 1.37 bran (N, P & K) 2015b) 2015b)

pg. 38

CONVERSION TO ELEMENTAL FORM Fertilizer inputs are expressed in elemental form for nitrogen (N) but in the oxide form for phosphorus (P2O5) and potassium (K2O). For this study, nutrients are expressed in actual elemental form (such as in Table 4.2, 4.2). Therefore, to convert P2O5 to P, multiply by 0.44. To convert K2O to K, multiply by 0.83.

To calculate the macronutrient amount in fodder, some special steps must be taken: 1. Sum inputs per each individual farm. 2. Multiply each sum by 180 days or 25 weeks (of the dry season) based on whether input was reported daily or weekly. Dry season in Ethiopia is typically from December to May (6 months) and fodder is supplied in this period. 3. Convert from local unit (Appendix 7.1) to kg. 4. Convert to % dry matter (DM). 5. Multiply by %N/P/K. 6. Average macronutrient amount to calculate mean (±SD) for each representative farm.

pg. 39

4.4.4 MACRONUTRIENT OUTPUT

The nutrient content and %DM of all outputs is available in Appendix 7.2. Each component has several outputs (Table 4.4). Outputs can have different functions. Typically, kocho, bula, barley and maize are consumed. Coffee, cabbage, onion, khat, milk, eggs, chicken, goat, sheep and cattle tend to be sold. Only in rare instances is whole livestock consumed. Enset leaves and grasses from enset, coffee and khat fields are used for livestock fodder.

Table 4.4: Output process and their respective outputs.

Output Code and Output process nutrients

Removal in Kocho, bula, enset leaves, maize grain, barley grain, OUT1 harvested cabbage, onion leaf and root, coffee berry, coffee bean, (N, P & K) product dwarf khat leaves and twigs, tall khat leaves and twigs

Removal in OUT2 Maize stover, barley straw left over on fields crop residue (N, P & K)

Livestock OUT3 Milk*, eggs, chicken (1.3 kg), goat (30 kg), sheep (30 output sold (N, P & K) kg), cattle (500 kg) that are sold products off-farm Livestock OUT4 Chicken (1.3 kg), goat (30 kg), sheep (30 kg), cattle (500 household (N, P & K) kg) that are consumed by household consumption Removal in Kocho, bula, enset leaves, maize grain, barley grain, harvested OUT5 cabbage, onion leaf and root, coffee berry, coffee bean, product sold (N, P & K) dwarf khat leaves and twigs, tall khat leaves and twigs off-farm that are sold products

* butter is included in this output

Note: For livestock, if the whole animal is sold the whole animal’s nutrient content is accounted for. Assumed body weights are listed in Table 4.3.

pg. 40

The formula used to extract macronutrient amount from these outputs is presented below (Equation 4.3):

For each individual farm’s (not representative farm) output:

`&a = b ∗ %cd ∗ # (Equation 4.3)

`&a e = f

Where `&a = g"hij#&aik(#a "gj&#a jl ("hℎ k#$kmk$&"n l"ig (kg/farm/yr) b = l"ig(i − i(qjia($ j&aq&a jl ("hℎ k#$kmk$&"n l"ig %cd = q(ih(#a"r( $iQ g"aa(i # = #&aik(#a hj#a(#a % jl j&aq&a e = "m(i"r( g"hij#&aik(#a "gj&#a lji i(qi('(#a"akm( l"ig (sr/l"ig/Qi) f = #&gR(i jl k#$kmk$&"n l"ig' tkaℎk# i(qi('(#a"akm( l"ig

Once e is established for all outputs, the sum (∑) can be taken for each component (Equation 4.4).

(Equation 4.4) uvf = w= =

Equation 4.4 is repeated for each macronutrient (N/P/K).

4.4.5 THE HARVEST INDEX

Nutrient removal in crop residue (OUT3) from maize and barley was calculated using the harvest index (HI). The harvest index is defined as the kg of grain divided by the total kg of above ground biomass (stover/straw plus grain).

x"im('a k#$(P = sr jl ri"k# / (sr jl 'ajm(i/'ai"t + sr jl ri"k#)

(Equation 4.5)

The HI used for barley was 0.39 and the HI used for maize was 0.52 (Mellisse et al., in prep.).

4.4.6 THE ENSET EXCEPTION

Determining the macronutrient content from enset requires particular attention. For perennial cash crops coffee and khat, the yield is harvested two to four times a year. Annual cereals and vegetables are annual crops which perform their entire life cycle from seed to flower within one growing season.

pg. 41

Enset is an exception. Within the home garden, enset is the only crop which is not harvested each year. In fact, as the primary subsistence crop and staple food, the enset harvest is dependent on household demand. Only some of the available enset plants are harvested every year, and four possible outputs are produced: kocho, bula, fibre and leaves. However, fibre was excluded from this study as it scarcely contains nutrients and enset leaves are accounted for as internal fodder. That leaves kocho and bula. The nutrient content for these outputs are listed in the table below (Table 4.5). Enset outputs are typically recorded in chinet, which equals 50 kg. First, outputs were converted into kilograms and dry matter. Second, DM (kg) is taken over the allocated enset land area (ha) to get the harvested yield in DM (kg) per ha per year. Third, the yield (DM kg/farm/yr) was multiplied by the nutrient content of the respective output (%DM output). This was the result per individual farm and it was repeated for all individual farms, for each macronutrient. The calculation was performed at component and farm level. For component level, all harvested output (OUT1) was taken into account. For the farm level, only the harvested output that is sold (OUT5) is taken into account. A mean (±SD) was taken for each macronutrient for both the component and farm level, for each representative farm.

P = Q ∗ 50 sr N = P ∗ %cd > R = . h = R ∗ f/w/z hj#a(#a k# (#'(a j&aq&a (%cd (#'(a j&aq&a)

h e = f

Where P = "gj&#a jl li('ℎ (#'(a j&aq&a (sr) (Equation 4.6) Q = "gj&#a jl li('ℎ (#'(a j&aq&a (k# hℎk#(a = 50 sr) N = "gj&#a jl (#'(a j&aq&a k# cd (sr) " = n"#$ "i(" "nnjh"a($ aj (#'(a h&nakm"akj# (ℎ") R = "gj&#a jl cd (#'(a j&aq&a (sr/l"ig/Qi) h = g"hij#&aik(#a "gj&#a jl (#'(a j&aq&a (sr/l"ig/Qi) e = "m(i"r( g"hij#&aik(#a "gj&#a jl (#'(a j&aq&a (sr/l"ig/Qi) f = #&gR(i jl k#$kmk$&"n l"ig' tkaℎk# ("hℎ i(qi('(#a"akm( l"ig

Table 4.5 Enset output dry matter content (%DM) and nutrient contents (mean ±SD) (HU Agricultural College Soil Lab, 2015; Wondo Genet College Soil Lab, 2015).

Output %DM N content (%) P content (%) K content (%) Kocho 31.15 1.14 (±0.67) 0.15 (±0.02) 0.63 (±0.25) Bula 53.69 0.99 (±0.05) 0.27 (±0.07) 0.46 (±0.14)

pg. 42

4.4.7 COMPONENT LEVEL AND FARM LEVEL MACRONUTRIENT BALANCE

After macronutrient amounts were calculated from inputs and outputs, the two were balanced in at the component level and farm level. To indicate either a nutrient surplus or deficit for all macronutrients, this formula was used (Equation 4.6):

d"hij#&aik(#a R"n"#h( = #&aik(#a k#q&a – #&aik(#a j&aq&a (4.7)

Note: A nutrient surplus = inputs > outputs and a nutrient deficit = outputs > inputs.

To finish, nutrient flow diagrams for each representative farm were produced (Figure 4.3). The diagram presents the inflows and outflows that are accounted for at the farm level. All nutrient flows are determined in kg/farm/yr.

pg. 43

Where fh = g"hij#&aik(#a (fwz) hj#a(#a (sr/l"ig/Qi) e = n"#$ "i(" "nnjh"a($ aj hijq h&nakm"akj# (ℎ") l = #&gR(i jl l"ig' (PℎkRkak#r hjgqj#(#a jl k#a(i('a a = aja"n #&gR(i jl l"ig' tkaℎk# ℎjg( r"i$(# aQq(

Figure 4.3: A general nutrient flow diagram of a home garden system. The black dashed line denotes the component level boundary of the farm. Inflows and outflows outside the boundary represent those at the farm level. Thin, gray dashed lines denote relationships excluded from the study. Labels in italic signify factors not quantified, for which it was still worth identifying their place within the system.

pg. 44

4.5 ETHICAL CONSIDERATIONS

Methods for this research are heavily reliant on detailed survey data from 63 home gardens. The data extracted from these surveys must be treated with care and confidentiality. During the survey process, the participants were fully informed of the aims of the survey and consent was obtained to participate. Although survey results have been translated and entered into the data set, distinguishing factors of farmers may be present. Therefore, discretion is taken by never leaving the data set unattended and treating farmer’s identifying information with the utmost confidentiality.

Issues of privacy are especially important in light of khat cultivation. Khat’s high economic price increasingly attracts thieves, especially at harvest time. It is not uncommon for khat fields to be guarded 24 hours a day, at and just prior to time of harvest. Although khat cultivation, sale and use are legal in Ethiopia, it remains a banned substance in most of the world. Research on MNUE of khat-based cultivation may be seen as a hindrance to economic profitability by farmers. To avoid emotional distress of khat farmers and to protect own personal safety, I only visited home gardens under the guidance of Drs. Mellisse. Throughout his PhD research, Drs. Mellisse has established long-standing and trusting relationships with farmers. These bonds are crucial to the success of my and Drs. Mellisse’s research and should never be jeopardized.

There are also ethical considerations for data collection. This study has little control over the ethical considerations of the survey design and execution, which occurred in 2013. However, within this study, more data was collected while carrying out the outlier accuracy check. When visiting home gardens and conversing with farmers (albeit via a translator), an ethical duty exists to respect each individual participant’s autonomy. As well, although some farmers participated in the survey in 2013, they may not have the same inclinations in 2015. An ethical duty also exists to resist soliciting and respect this decision.

While carrying out the data analysis, ethics also play a role. For example, all results whether positive, significant or negative should be reported. Failing to report negative findings is misconduct and will severely weaken the study’s conclusions. Also, the data I collected to use should be well-preserved for potential future research. Changing the hypothesis of the paper, using other research’s words or data and/or editing or producing false data are major misconducts and are to be avoided at all costs.

Overall, this study aims to improve farmer and academic knowledge on the macronutrient inflows, outflows and balances across five representative farms. However, the role of ethical considerations in a study of this nature is not to be understated. By taking ethical considerations for data collection and analysis into account, this study can successfully realize its objectives.

pg. 45

5. RESULTS

In this chapter the results are presented. First, the representative farms, farm size and livestock population results are shown. Second, nutrient flows at the component level, followed by the farm level, are presented in figures, elaborated in tables and illustrated in a nutrient flow diagram for each home garden type. Third, results are converted to per hectare basis to aid discussion and comparative analysis.

5.1 THE REPRESENTATIVE FARMS

Land use of representative farms expressed in area shares is presented in Figure 5.1. The representative farm for an enset-based home garden, based on nine farms, was 1.10 ha. Land use comprised two-thirds (66%) enset cultivation, a quarter (28%) ACV and a small area (6%) of grazing land. Faba beans, a crop excluded from this study, and coffee, which was only present on 3 of 9 farms, were unaccounted for. The enset-based representative farm did not produce any bula, their only enset outputs were kocho and grass from enset fields. The average TLU for this representative farm was 0.43 and there was little household consumption of livestock, averaging at 0.18, 0.04 and 0.02 kg/farm/yr of N, P and K, respectively.

The representative enset-coffee home garden, based on 18 farms was 1.21 ha. This is the only home garden system with intercropping of enset and coffee (23%) and the only one to show traditionally combined production of the food crop enset (36%) and the cash crop coffee (24%), together taking up 83% of the land area. The remaining land was allocated to ACV (10%) and grazing land (7%). The area unaccounted for was covered in khat, but only on 8 out of 18 farms, excluding it from the analysis. The average TLU for an enset-coffee homegarden was 0.46 with household livestock consumption comparable to that in an enset-based homegarden.

The representative enset-cereal-vegetable home garden, based on nine farms, was 1.15 ha. It cultivated ACV (49%), enset (32%) and grazing land (19%). No area or crop was unaccounted for in this home garden. Enset-cereal-vegetable home gardens did not cultivate any maize, only barley, cabbage and onion. The average TLU was 0.64 with identical household livestock consumption as an enset-based system.

The representative enset-livestock home garden, based on nine farms, was 1.12 ha. Grazing land is represented by 39% in the enset-livestock system, more than in any other representative farm. The remaining area is cultivated with enset (34%), khat (20%) and ACV (7%). No area or crop was unaccounted for in this home garden. The average TLU was 0.77, more than in any other home garden. Household livestock consumption was also higher than in any other system, with 1.40, 0.09 and 0.11 kg/farm/yr of N, P and K, respectively.

pg. 46

Figure 5.1: Land use of representative farms expressed in area shares (%). pg. 47

The representative khat-based home garden, based on 18 farms, was 1.03 ha. Khat cultivation covered almost half of the area (45%), with relatively equal shares of enset (24%), ACV (16%) and grazing land (15%). Land use unaccounted for was 8%, likely attributed to potatoes, a crop excluded from this analysis, and coffee, which was only present on 8 of 18 farms. The average TLU was 0.62, there was no livestock consumed by khat-based households.

5.2 FARM SIZE

The average farm size per household was not substantially different amongst the five home garden types. The largest (1.21 ha) farm size was observed for the enset-coffee system, while the smallest (1.03 ha) was the khat-based system. The farm size in the other three home gardens were between these two home garden types. The accuracy of farmer-reported farm size was crosschecked by measuring the area of each land use type with Global Positioning System (GPS) devices (Figure 5.2). This was done for 24 of the 63 farms surveyed.

The scatter plot of GPS measured (independent variable) against farmer-reported (dependant variable) had a strong relationship, with a range of coefficient of determination (R2) values of 0.6 to 0.9, except for ACV (R2 = 0.1) (Table 5.1). The relationship was stronger for perennial crops (enset, coffee, khat) and permanent land use of grazing land than annual-based land use types, such as ACV. Of the permanent crops, grazing land had strongest relationship with a coefficient of determination (R2) value of close to one. Across all home garden systems, grazing land is ubiquitous and its generally small land allocation is constant. For enset, as the staple crop, also reported and measured areas were in reasonable agreement with an R2 value of 0.82. Any data point above the y=x linear reference line represents under-reporting of land size, any data below the line represents over-reporting

Overall, 12 of 24 farmers over-reported their total land size, the other half under-reported. However, when ACV were excluded due to their significant divergence (R2 = 0.11), 15 of 24 farmers were over-reporting with on average 0.2 ha. On the contrary, when farmers under- reported they did so by an average of 0.1 ha.

Table 5.1: Regression equations and R-Squared values for GPS measured land size (ha, x-axis) vs. farmer reported land size (ha, y-axis) by component.

Farm component Equations of regression lines R-Squared (R2) Enset y = 1.30x - 0.03 0.82 Coffee y = 1.17x + 0.05 0.69 ECI y = 0.75x - 0.04 0.57 ACV y = 0.44x + 0.17 0.11 Khat y = 0.90x + 0.09 0.57 Grazing land y = 0.82x + 0.05 0.96

Note: ECI = Enset + coffee intercrop, ACV = Annual cereals and vegetables

pg. 48

Figure 5.2: GPS measured land size (ha) vs. farmer reported land size (ha) by farm component.

pg. 49

5.3 LIVESTOCK POPULATION

Livestock is an essential component of the home garden system. Average TLU was calculated for each representative farm and plotted against area share of grazing land (ha) for each representative farm (Figure 5.3). The data points are close to the linear trend line with an R2 value of 0.90. A positive correlation also exists between TLU and area share of grazing land. When TLU increases, the area share allocated to grazing land grows. A logical correlation as higher TLU characteristically requires greater share of grazing land.

Enset-based Enset-coffee Enset-cereal-vegetable Enset-livestock Khat-based

Figure 5.3: Area share of grazing land (ha) in each representative farm vs. average Tropical Livestock Unit (TLU) for each representative farm, by home garden type.

pg. 50

5.4 COMPONENT LEVEL NUTRIENT BALANCE ASSESSMENT

The balances were calculated by aggregating the inflows and outflows across all farm components (enset, coffee, enset + coffee intercropping, annual cereals and vegetables, khat and livestock) within the representative farms. All nutrient amounts are reported in kilogram per farm per year (kg/farm/yr). This section also presents nutrient flow diagrams for each representative farm. 5.4.1 ENSET-BASED

Figure 5.4 shows the component level nutrient inflows, outflows and balances for an enset- based representative farm. In an enset-based system, organic matter (IN2) was the primary source of NPK to enset. It supplies 108 kg of N, 4 kg of P and 38 kg of K. The major input source for ACV was mineral fertilizer (IN1), supplying 14 kg N and 4 kg P. Internal fodder (IN3) inputs to livestock component were lowest amongst all representative farms with 10 kg N, 4 kg P and 29 kg K. Enset-based farms also have the lowest TLU (0.43) indicating either a decreased demand for their own internal fodder use or the practice of feeding its internal fodder to temporary livestock. This practice would not be reported on the input/output survey but could explain the high organic matter available to these typically poorer-farmers. External fodder was very small, under 0.5 kg NPK.

Outflows were primarily through removal in harvested products (OUT1), particularly enset output (kocho) from enset component and ACV output (barley). Crop residues were so small, N, P and K were all under 0.5 kg NPK from ACV component. Removal through whole livestock and livestock products (OUT3) was 2 kg N from livestock component. Household livestock consumption (OUT4) was also very small, under 0.5 kg NPK.

Nutrient balances in the components of enset-based farms were nearly all positive with the exception of a small negative K balance in the ACV component. N balances on enset fields were highly positive, likely attributed to their exorbitant organic matter inputs from their “guest” livestock. P balances were slightly positive or neutral throughout all three farm components. The livestock K balance was positive, but only due to its high K inputs from internal fodder and lack of K in its livestock outputs. Figure 5.5 presents the nutrient flows of the enset-based system.

pg. 51

Figure 5.4: Component level nutrient inflows, outflows and balances for N, P and K (kg/farm/yr) for an enset-based representative farm.

pg. 52

Figure 5.5: Nutrient flows (kg/farm/yr) that influences the partial nutrient balance of an enset-based system. The asterisk (*) after organic matter (IN2) denotes that this input likely came from an external source, as 0.43 TLU could not have produced this much compost. The black dashed line denotes the component level boundary of the farm. Inflows and outflows outside the boundary represent those at the farm level. The gray dashed lines denote relationships which were excluded from the study. Labels in italic signify factors not quantified, for which it was still worth identifying their place within the system.

pg. 53

5.4.2 ENSET-COFFEE

Figure 5.6 shows the component level nutrient inflows, outflows and balances for an enset- coffee representative farm. In an enset-coffee system, there were lower mineral fertilizer (IN1) inputs than in an enset-based system and it was only applied to ACV fields. These inflows were calculated to be 6 kg N and 2 kg P in the ACV component. It is also the smallest influx of mineral fertilizers amongst all representative farms. The other components (enset, coffee and ECI) relied on organic matter. The most was applied to in the coffee component, likely prioritized as coffee is the primary cash crop in this representative farm. Farmers applied 11 kg N and 4 kg K to the coffee component. P amount on these fields was under 0.5 kg. The livestock component received 17 kg N, 5 kg P and 46 kg K from internal fodder (IN3). Surprisingly, enset-coffee farms had the largest input of external fodder (IN4) amongst all representative farms. Inputs from wheat bran and sugarcane tops combined were 5 kg N, 1 kg P and 8 kg K to the livestock component.

The amount of NPK removed from harvested products in enset-coffee farms were relatively equally distributed amongst the enset, coffee and ECI components. Each component removed 5 kg N and 1 kg P. K varied more, with 3 kg, 9 kg and 6 kg removed from enset, coffee and ECI fields, respectively. Nutrient removals from the ACV component were 2 kg N, 1 kg P and 1 kg K. The enset-coffee representative farm was the only one with coffee and enset + coffee intercropping and therefore the only home garden to display NPK influx and removal from these components. Nutrient removals from crop residues were negligible in the ACV component and livestock outputs were slight with just 2 kg N in the livestock component.

Across all components N balances were positive. The highest was in the livestock component with an N surplus of 20 kg N. For enset, coffee and ECI, P and K balances were all negative with the largest on coffee fields (- 5 kg K). Although the ACV component had a positive P balance, its K balance was also negative. In an enset-coffee representative farm, ACV fields received only mineral fertilizers which do not give a source of K. On the other hand, enset, coffee and ECI farms do not receive enough organic matter and consequently inadequate amounts of K. Livestock balances were positive, almost matching NPK inputs as there is so little output from the livestock component. Figure 5.7 presents the nutrient flows of the enset-coffee system.

pg. 54

Figure 5.6: Component level nutrient inflows, outflows and balances for N, P and K (kg/farm/yr) for an enset-coffee representative farm.

pg. 55

Figure 5.7: Nutrient flows (kg/farm/yr) that influences the partial nutrient balance of an enset-coffee system. The black dashed line denotes the component level boundary of the farm. Inflows and outflows outside the boundary represent those at the farm level. The gray dashed lines denote relationships which were excluded from the study. Labels in italic signify factors not quantified, for which it was still worth identifying their place within the system. pg. 56

5.4.3 ENSET-CEREAL-VEGETABLE

Figure 5.8 presents the component level inflows, outflows and balances for an enset-cereal- vegetable representative farm. In an enset-cereal-vegetable system, the enset component had organic matter inputs of 23 kg N, 1 kg P and 8 kg N. The ACV component had a combination of organic matter and mineral fertilizer applied, although it was primarily supplied with mineral fertilizers, with 16 kg N and 8 kg P. A small amount of K (2 kg) was supplied with the organic matter in the ACV component. Across representative farms, both enset-based and enset-cereal- vegetable representative farms had the lowest influx of nutrients via internal fodder and no input from external fodder. NPK input to the livestock component was 19 kg, 6 kg and 52 kg, respectively.

The ACV component was the largest source of nutrient depletion in an enset-cereal-vegetable system by the nutrient removals of harvested ACV products, valued at 19 kg N, 7 kg P and 14 kg K. A signal the system is appropriately named after its considerable annual cereal and vegetable production. The enset-cereal-vegetable system was the only one amongst all representative farms to have any nutrients removed via crop residues; these were valued at 1 kg N and 1 kg K. These crop residues are either added to the compost pile, or used as livestock feed.

A small negative P balance existed in the enset component. Larger negative P and K balances existed in the ACV component. P was depleted by 7 kg and K was depleted 9 kg. Livestock balances were very positive in the livestock component. Figure 5.9 presents the nutrient flows of the enset-cereal-vegetable system.

pg. 57

Figure 5.8: Component level nutrient inflows, outflows and balances for N, P and K (kg/farm/yr) for an enset-cereal-vegetable representative farm.

pg. 58

Figure 5.9: Nutrient flows (kg/farm/yr) that influences the partial nutrient balance of an enset-cereal-vegetable system. The black dashed line denotes the component level boundary of the farm. Inflows and outflows outside the boundary represent those at the farm level. The gray dashed lines denote relationships which were excluded from the study. Labels in italic signify factors not quantified, for which it was still worth identifying their place within the system. pg. 59

5.4.4 ENSET-LIVESTOCK

Figure 5.10 presents the component level inflows, outflows and balances for an enset-livestock representative farm. Enset-livestock farms have the largest land use allocated to livestock (grazing land = 0.39 ha) and the largest TLU (0.77). However, they have lower influx of external fodder than enset-coffee farms, with 1 kg N and 1 kg from wheat bran and sugarcane top. Inputs of sugarcane tops were only observed in 1 of 9 surveyed farms within the enset-livestock representative farm. Internal fodder inputs were highest amongst all representative farms, but only slightly. Influxes of nutrients from enset leaves and grasses were calculated at 22 kg N, 8 kg P and 66 kg K to the livestock component. The enset-livestock representative farm is one of two which allocates land to khat, the other being the khat-based system. This change causes a spike in mineral fertilizer. The khat component in enset-livestock farms received 25 kg N and 4 kg P. The ACV component received substantially lower mineral fertilizers with 1 kg N. The enset component relied on organic matter with 13 kg N and 4 kg K.

Compared to enset-cereal-vegetable farms, enset-livestock systems saw a drastic drop of macronutrients removed via harvested ACV products with 1 kg N, 1 kg P and 1 kg K, removed. Enset-livestock farms produce very little ACV output. The khat component depleted 7 kg N, 2 kg P and 8 kg K. Livestock outputs rose substantially with 17 kg N, 4 kg P and 3 kg P removed via either whole livestock or livestock products. Household livestock consumption removed only 1 kg N but was the only representative farm to do so. Dismal figures for household livestock consumption across representative farms suggest livestock consumption is of little relevance in all of these systems and may only occur for special occasions. Harvested products from the enset component removed 9 kg N, 1 kg P and 7 kg K.

The introduction of khat, which receives only mineral fertilizers in an enset-livestock representative farm, has introduced K deficiencies on these fields. In this assessment, khat is depleting soil of -8 kg K per year. The enset component also observe negative balances for both P and K; likely a result of receiving too little organic matter. The ACV component receives so little organic matter, only 1 kg N is applied and virtually no nutrients are removed. This leaves a negative, but small, N balance. Figure 5.11 presents the nutrient flows of the enset-cereal- vegetable system.

pg. 60

Figure 5.10: Component level nutrient inflows, outflows and balances for N, P and K (kg/farm/yr) for an enset-livestock representative farm.

pg. 61

Figure 5.11: Nutrient flows (kg/farm/yr) that influences the partial nutrient balance of an enset-livestock system. The black dashed line denotes the component level boundary of the farm. Inflows and outflows outside the boundary represent those at the farm level. The gray dashed lines denote relationships which were excluded from the study. Labels in italic signify factors not quantified, for which it was still worth identifying their place within the system. pg. 62

5.4.5 KHAT-BASED

Figure 5.12 presents the component level inflows, outflows and balances for a khat-based representative farm. The influxes of mineral fertilizers in this system are the highest amongst all representative farms. In the khat component, 83 kg N and 13 kg P was applied to khat fields. Mineral fertilizers were also applied to ACV fields with 9 kg N and 4 kg P. Organic matter was applied to the enset component with 12 kg N and 4 kg K added. Internal fodder was on the lower spectrum, second to last amongst all representative farms. They were valued at 14 kg N, 4 kg P and 39 kg K. External fodder were valued at 4 kg N, 1 kg P and 3 kg K, just behind enset- coffee systems, in the livestock component.

The khat-based representative farm was aptly named for its 10 kg N, 3 g P and 10 kg K removal from the khat component, the largest amongst all farm types. The system removed 4 kg N, 3 kg P and 3 kg K from its ACV component and 7 kg N, 1 kg P and 3 kg K from its enset component. Nutrient removal from whole livestock and livestock products was 8 kg N, 2 kg P and 2 kg K and household livestock consumption was non-existent in the livestock component.

The K balance for the ACV component was slightly negative. The N balance for the khat component was very positive indicating excessive mineral fertilizer application. Khat fields had a severe K deficiency of -10 kg. The remaining balances (enset component, livestock component) were mainly positive with the K balance in the livestock component being very high. Figure 5.13 presents the nutrient flows of the enset-cereal-vegetable system.

Table 5.2 elaborates on the mean (±SD) component level NPK inflows (kg/farm/yr) by farm component for each representative farm. The table separates NPK inputs by input source: mineral fertilizer (IN1) is DAP and/or urea, organic matter (IN2) is compost, internal livestock fodder (IN3) is grass or enset leaves and external livestock fodder (IN4) is sugarcane top and/or wheat bran. A code is assigned to each farm component and its respective output(s). In the brackets behind the output function code (OUT) is that output’s product. For instance, OUT1 is split into kocho and bula products. Table 5.3 does the same for component level NPK outflows (kg/farm/yr) by farm component for each representative farm.

pg. 63

Figure 5.12: Component level nutrient inflows, outflows and balances for N, P and K (kg/farm/yr) for a khat-based representative farm.

pg. 64

Figure 5.13: Nutrient flows (kg/farm/yr) that influences the nutrient balance of a khat-based system. The black dashed line denotes the component level boundary of the farm. Inflows and outflows outside the boundary represent those at the farm level. The gray dashed lines denote relationships which were excluded from the study. Labels in italic signify factors not quantified, for which it was still worth identifying their place within the system. pg. 65

Table 5.2: Component level macronutrient inflows (kg/farm/yr) from mineral fertilizers (IN1), organic matter (IN2), internal fodder (IN3) and external fodder (IN4) (mean ±SD) by farm component, across five representative farms.

Input Farm type Farm type Farm type Farm type Farm type Component function Enset-based Enset-coffee Enset-cereal-vegetable Enset-livestock Khat-based (IN) N n P n K n N n P n K n N n P n K n N n P n K n N n P n K n IN2 108 (±72) 9 4 (±3) 9 38 (±25) 9 7 (±6) 16 0 (±0) 16 3 (±2) 16 23 (±26) 9 1 (±1) 9 8 (±9) 9 13 (±14) 9 0 (±1) 9 4 (±5) 9 12 (±10) 18 0 (±0) 18 4 (±3) 18 Enset Total 108 4 38 7 0 3 23 1 8 13 0 4 12 0 4 IN2 11 (±10) 14 0 (±0) 14 4 (±3) 14 Coffee Total 11 0 4 IN2 6 (±6) 10 0 (±0) 10 2 (±2) 10 ECI Total 6 0 2 IN1 (DAP) 4 (±3) 8 4 (±3) 8 2 (±2) 11 2 (±2) 11 7 (±5) 21 8 (±6) 21 3 (±2) 14 4 (±2) 14 IN1 (urea) 10 (±8) 8 5 (±5) 11 9 (±12) 21 5 (±6) 14 ACV IN2 0 (±1) 8 0 (±0) 8 0 (±0) 8 7 (±8) 21 0 (±0) 21 2 (±3) 21 1 (±1) 7 0 (±0) 7 0 (±0) 7 Total 14 4 0 6 2 23 8 2 1 0 0 9 4 0 IN1 (DAP) 4 (±2) 8 4 (±2) 8 0 11 (±9) 17 13 (±10) 17 IN1 (urea) 21 (±11) 8 0 0 71 (±35) 17 Khat IN2 0 0 0 0 (±0) 17 0 (±0) 17 0 (±0) 17 Total 25 4 0 83 13 0 IN3 (grass) 2 (±1) 6 1 (±0) 6 2 (±2) 6 5 (±5) 16 1 (±1) 16 6 (±6) 16 7 (±3) 9 2 (±1) 9 9 (±4) 9 5 (±2) 9 2 (±1) 9 7 (±3) 9 4 (±2) 16 1 (±1) 16 5 (±3) 16 IN3 (EL) 8 (±5) 8 3 (±2) 8 27 (±19) 8 12 (±9) 15 4 (±3) 15 40 (±30) 15 12 (±7) 9 4 (±2) 9 43 (±24) 9 17 (±5) 9 6 (±2) 9 59 (±18) 9 10 (±5) 18 3 (±2) 18 34 (±19) 18 Livestock IN4 (SCT) 3 (±4) 7 0 (±1) 7 7 (±10) 7 0 (n/a) 1 0 (n/a) 1 1 (n/a) 1 1 (±1) 8 0 (±0) 8 2 (±2) 8 IN4 (WB) 0 (±0) 3 0 (±0) 3 0 (±0) 3 2 (±2) 16 1 (±1) 16 1 (±1) 16 0 (±1) 5 0 (±0) 5 0 (±0) 5 1 (±1) 6 0 (±0) 6 0 (±0) 6 3 (±3) 10 1 (±1) 10 1 (±2) 10 Total 10 4 29 22 6 54 19 6 52 23 8 67 17 5 42 TSN 132 12 67 52 8 62 65 15 62 62 13 72 120 22 46

Note: EL = enset leaves, SCT = sugarcane top and WB = wheat bran. N/A = not applicable, written if there was only one observation.

pg. 66

Table 5.3: Component level macronutrient outflows (kg/farm/yr) from removal in harvested products (OUT1), removal in crop residues (OUT2), whole livestock and livestock products sold off-farm (OUT3) and household livestock consumption (OUT4) (mean ±SD) by farm component, across five representative farms.

Output Farm type Farm type Farm type Farm type Farm type Component function Enset-based Enset-coffee Enset-cereal-vegetable Enset-livestock Khat-based (OUT) N n P n K n N n P n K n N n P n K n N n P n K n N n P n K n OUT1 (kocho) 24 (±19) 9 3 (±3) 9 13 (±11) 9 5 (±5) 16 1 (±1) 16 3 (±3) 16 10 (±5) 9 1 (±1) 9 6 (±3) 9 9 (±6) 9 1 (±1) 9 5 (±3) 9 6 (±4) 18 1 (±1) 18 0 (±0) 18 OUT1 (bula) 0 (±0) 16 0 (±0) 16 0 (±0) 16 0 (±0) 6 0 (±0) 6 0 (±0) 6 0 (±0) 9 0 (±0) 9 0 (±0) 9 0 (±0) 17 0 (±0) 17 0 (±0) 17 Enset OUT1 (leaves) 0 (±0) 3 0 (±0) 3 0 (±0) 3 0 (±0) 9 0 (±0) 9 2 (±1) 9 1 (±1) 13 0 (±0) 13 3 (±3) 13 Total 24 3 13 5 1 3 11 1 6 9 1 7 7 1 3 OUT1 (coffee berry) 4 (±2) 18 1 (±1) 17 8 (±5) 18 Coffee OUT1 (coffee bean) 1 (±1) 17 0 (±0) 17 1 (±1) 17 Total 5 1 9 OUT1 (enset) 3 - - 0 - - 1 - - ECI OUT1 (coffee) 2 - - 0 - - 4 - - Total 5 1 6 OUT1 (barley) 1 (±1) 4 1 (±0) 4 1 (±1) 4 0 (n/a) 1 0 (n/a) 1 0 (n/a) 1 2 (±1) 8 1 (±1) 8 1 (±1) 8 0 (n/a) 1 0 (n/a) 1 0 (n/a) 1 1 (±2) 5 1 (±1) 5 1 (±1) 5 OUT1 (maize) 1 (±1) 2 0 (±0) 2 0 (±0) 2 1 2 9 1 (±1) 9 1 (±1) 9 0 (±1) 3 0 (±1) 3 0 (±0) 3 3 (±4) 10 1 (±2) 10 1 (±2) 10 OUT1 (cabbage) 0 (±1) 2 0 (±1) 2 0 (±1) 2 4 (±2) 5 5 (±3) 5 3 (±2) 5 0 (n/a) 1 0 (n/a) 1 0 (n/a) 1 ACV OUT1 (onion) 0 (n/a) 1 0 (n/a) 1 0 (n/a) 1 13 (±8) 8 2 (±1) 8 9 (±5) 8 OUT2 (barley) 0 (±0) 4 0 (±0) 4 0 (±0) 4 0 (n/a) 1 0 (n/a) 1 0 (n/a) 1 1 (±0) 8 0 (±0) 8 1 (±0) 8 0 (n/a) 1 0 (n/a) 1 0 (n/a) 1 0 (±0) 5 0 (±0) 5 0 (±1) 5 OUT2 (maize) 0 (±0) 2 0 (±0) 2 0 (±0) 2 0 (±0) 9 0 (±1) 9 0 (±0) 9 0 (±0) 3 0 (±0) 3 0 (±0) 3 0 (±1) 10 1 (±2) 10 1 (±1) 10 Total 2 1 1 2 1 1 19 7 14 1 1 1 4 3 3 OUT1 (leaves, twigs) 7 (±5) 8 2 (±2) 8 8 (±5) 8 10 (±6) 18 3 (±2) 18 10 (±7) 18 Khat Total 7 2 8 10 3 10 OUT3 2 (±1) 8 0 (±0) 8 0 (±0) 8 2 (±1) 8 0 (±0) 8 0 (±0) 8 6 (±3) 8 2 (±1) 8 1 (±1) 8 17 (±9) 9 4 (±2) 9 3 (±1) 9 8 (±4) 18 2 (±1) 18 2 (±1) 18 Livestock OUT4 0 (±1) 2 0 (±1) 2 0 (±1) 2 0 (n/a) 1 0 (n/a) 1 0 (n/a) 1 0 (±1) 4 0 (±0) 4 0 (±0) 4 1 (±1) 3 0 (±0) 3 0 (±0) 3 Total 2 0 0 2 0 0 6 2 1 18 4 3 8 2 2 TSN 28 5 15 18 4 19 36 11 21 36 8 18 28 9 18 Note: N/A = not applicable, written if there was only one observation. ECI output reported as half of enset macronutrient output and half of coffee macronutrient output, due to data availability. This method could not produce a value for standard deviation and number of observations for the ECI component. This is denoted with a small dash.

pg. 67

5.5 FARM LEVEL NUTRIENT BALANCE ASSESSMENT

Figure 5.14 presents the farm level inflows, outflows and balances for NPK by representative farm type. For the farm level nutrient balance assessment only mineral fertilizers (IN1) and external fodder (IN4) inflows were taken into account. The outflows considered were nutrient removal in harvested products sold off-farm (OUT5) and whole livestock and livestock products sold off-farm (OUT3). At the farm level, N balances were most varied at the farm level. Those of enset-based and enset-coffee farms were slightly positive. The N balance for enset-cereal- vegetable farms was -8 kg N negative. However, Figure 5.9 reveals this is due to enset-cereal- vegetable farms high livestock output but lack of external fodder. Enset-livestock and khat- based farms have very high N balances due to their high mineral fertilizer inputs. The khat- based N balance is particularly high at 94 kg N.

P balances were neutral amongst all representative farms, except for the khat-based system. In this representative farm, P increased with 12 kg. This was attributed to khat-based propensity for high mineral fertilizer application. In this case, DAP was in excess as it is the only mineral fertilizer to supply P.

K balances were negative throughout all representative farms. K deficiencies in enset-based and enset-cereal-vegetable were especially noticeable. Figure 5.5 shows the nutrient output within enset harvested products, but no external inputs to enset fields. As such the K, which is sufficient from internal input, is deemed insufficient. In fact, enset-fields receive 22 kg K from organic matter in enset-based farms. K deficiency in enset-cereal-vegetable farms can be attributed to the large nutrient removal from harvested ACV products sold off-farm but only fertilized with mineral N and P fertilizers (Figure 5.9). Crops fertilized only with mineral fertilizers will face K deficiencies. Enset-coffee, enset-livestock and khat-based systems all had relatively similar K deficiencies at the farm level. The enset-coffee representative farm does not receive adequate organic matter to meet the amounts removed via harvested crops (Figure 5.7). Coffee especially is a K-rich crop. Coffee berries have 3.19% K and coffee beans have 2.16% K (Hawassa University Agricultural College Soil Laboratory, 2015; Wondo Genet College Soil Laboratory, 2015). Enset-livestock and khat-based farms had K deficiencies as K was never applied through external inputs.

K scarcity in internal input-reliant farms is exaggerated in the farm level analysis as the assessment does not take K inputs from organic matter and internal fodder into account. For instance, the high positive K balances from internal fodder (especially enset leaves with 4.60% K content (4.60% K) to livestock that is seen across all representative farms is concealed from farm balances. Table 5.4 elaborates on the mean (±SD) farm level macronutrient inflows (kg/farm/yr) and total sum of nutrient (TSN) from mineral fertilizers (IN1) and external fodder (IN4) by farm component, across five representative farms. Table 5.5 does the same for farm level macronutrient outflows (kg/farm/yr) and total sum of nutrient (TSN) from removal in harvested products sold off-farm (OUT5) and whole livestock and livestock products sold off- farm (OUT3) by farm component, across five representative farms.

pg. 68

Figure 5.14: Farm level nutrient inflows, outflows and balances for N, P and K (kg/farm/yr) across representative farms.

pg. 69

Table 5.4: Farm level macronutrient inflows (kg/farm/yr) from mineral fertilizers (IN1) and external fodder (IN4) (mean ±SD) by farm component, across five representative farms.

Input Farm type Farm type Farm type Farm type Farm type Component function Enset-based Enset-coffee Enset-cereal-vegetable Enset-livestock Khat-based (IN) N n P n K n N n P n K n N n P n K n N n P n K n N n P n K n IN1 (DAP) 4 (±3) 8 4 (±3) 8 2 (±2) 11 2 (±2) 11 7 (±5) 21 8 (±6) 21 3 (±2) 14 4 (±2) 14 ACV IN1 (urea) 10 (±8) 8 5 (±5) 11 9 (±12) 21 5 (±6) 14 Total 13 4 6 2 16 8 9 4 IN1 (DAP) 4 (±2) 8 4 (±2) 8 0 11 (±9) 17 13 (±10) 17 Khat IN1 (urea) 21 (±11) 8 0 0 71 (±35) 17 Total 25 4 82 13 IN4 (SCT) 3 (±4) 7 0 (±1) 7 7 (±10) 7 0 (n/a) 1 0 (n/a) 1 1 (n/a) 1 1 (±1) 8 0 (±0) 8 2 (±2) 8 Livestock IN4 (WB) 0 (±0) 3 0 (±0) 3 0 (±0) 3 2 (±2) 16 1 (±1) 16 1 (±1) 16 0 (±1) 5 0 (±0) 5 0 (±0) 5 1 (±1) 6 0 (±0) 6 0 (±0) 6 3 (±3) 10 1 (±1) 10 1 (±2) 10 Total 0 0 0 5 1 8 0 0 0 1 0 1 3 1 3

TSN 14 4 0 11 3 8 16 8 0 27 4 1 94 18 3

Note: SCT = sugarcane top and WB = wheat bran. N/A = not applicable, written if there was only one observation.

pg. 70

Table 5.5: Farm level macronutrient outflows (kg/farm/yr) from removal in harvested products sold off-farm (OUT5) and whole livestock and livestock products sold off-farm (OUT3) (mean ±SD) by farm component, across five representative farms.

Output Farm type Farm type Farm type Farm type Farm type Component function Enset-based Enset-coffee Enset-cereal-vegetable Enset-livestock Khat-based (OUT) N n P n K n N n P n K n N n P n K n N n P n K n N n P n K n Enset OUT5 (kocho) 5 (±7) 9 1 (±1) 9 3 (±4) 9 1 (±2) 10 0 (±0) 10 0 (±1) 10 2 (±1) 7 0 (±0) 7 1 (±1) 7 0 (±0) 9 0 (±0) 9 0 (±0) 9 1 (±0) 18 0 (±0) 18 1 (±1) 18 OUT5 (bula) 0 (±0) 2 0 (±0) 2 0 (±0) 2 0 (±0) 5 0 (±0) 5 0 (±0) 5 0 (±0) 6 0 (±0) 6 0 (±0) 6 0 (±0) 12 0 (±0) 12 0 (±0) 12 OUT5 (leaves) 0 (±0) 3 0 (±0) 3 0 (±0) 3 0 (±0) 9 0 (±0) 9 2 (±1) 9 1 (±1) 13 0 (±0) 13 3 (±3) 13 Total 5 1 3 1 0 0 2 0 1 0 0 0 1 0 1 Coffee OUT5 (coffee berry) 4 (±2) 18 1 (±0) 18 8 (±5) 18 OUT5 (coffee bean) 1 (±1) 14 0 (±0) 14 1 (±1) 14 Total 5 1 9 ECI OUT5 (enset) 1 - - 0 - - 0 - - OUT5 (coffee) 2 - - 0 - - 4 - - Total 3 0 4 ACV OUT5 (barley) 0 (±0) 2 0 (±0) 2 0 (±0) 2 0 (±1) 2 0 (±1) 2 0 (±1) 2 0 (n/a) 1 0 (n/a) 1 0 (n/a) 1 OUT5 (maize) 0 (±1) 9 0 (±0) 9 0 (±0) 9 1 (±2) 4 0 (±1) 4 0 (±1) 4 OUT5 (cabbage) 1 (±0) 2 1 (±0) 2 0 (±0) 2 3 (±3) 5 3 (±4) 5 3 (±2) 5 OUT5 (onion) 0 (n/a) 1 0 (n/a) 1 0 (n/a) 1 13 (±7) 8 2 (±1) 8 9 (±5) 8 Total 1 1 0 0 0 0 16 5 12 0 0 0 1 0 0 Khat OUT5 (leaves, twigs) 5 (±4) 8 2 (±1) 8 5 (±4) 8 10 (±6) 18 3 (±2) 18 10 (±7) 18 Total 5 2 5 10 3 10 Livestock OUT3 2 (±1) 8 0 (±0) 8 0 (±0) 8 2 (±1) 8 0 (±0) 8 0 (±0) 8 6 (±3) 8 2 (±1) 8 1 (±1) 8 17 (±9) 9 4 (±2) 9 3 (±1) 9 8 (±4) 18 2 (±1) 18 2 (±1) 18 Total 2 0 0 2 0 0 6 2 1 17 4 3 8 2 2 TSN 8 2 3 10 2 14 24 7 14 22 6 8 19 5 13 Note: N/A = not applicable, written if there was only one observation. ECI output reported as half of enset macronutrient output and half of coffee macronutrient output, due to data availability. This method could not produce a value for standard deviation and number of observations for the ECI component. This is denoted with a small dash.

pg. 71

5.6 RESULTS PER HECTARE

To ease comparative analysis with literature in the discussion, the results for crop components (enset, coffee, ECI, ACV and khat) originally reported in kg/farm/yr have been converted to kg/ha/yr. The livestock component was not converted to a per hectare basis because the inherent nature of livestock as an animal (and not a crop) does not allow this conversion. Table 5.6 shows component level macronutrient inflows, outflows and balances (kg/ha/yr). Table 5.7 shows farm level macronutrient inflows, outflows and balances (kg/ha/yr).

pg. 72

Table 5.6: Component level macronutrient inflows, outflows and balances (kg/ha/yr) by farm component, across five representative farms.

Farm type Farm type Farm type Farm type Farm type Component Enset-based Enset-coffee Enset-cereal-vegetable Enset-livestock Khat-based N FS N P FS P K FS K N FS N P FS P K FS K N FS N P FS P K FS K N FS N P FS P K FS K N FS N P FS P K FS K IN 108 0.72 150 4 0.72 5 38 0.72 53 7 0.45 16 0 0.45 1 3 0.45 6 23 0.37 61 1 0.37 2 8 0.37 21 13 0.38 34 0 0.38 1 4 0.38 12 12 0.25 47 0 0.25 2 4 0.25 16 Enset OUT 24 0.72 33 3 0.72 4 13 0.72 18 5 0.45 11 1 0.45 2 3 0.45 6 11 0.37 29 1 0.37 4 6 0.37 16 9 0.38 24 1 0.38 3 7 0.38 19 7 0.25 29 1 0.25 3 3 0.25 12 BAL 84 0.72 117 1 0.72 1 25 0.72 34 2 0.45 5 0 0.45 -1 0 0.45 -1 12 0.37 33 -1 0.37 -2 2 0.37 6 4 0.38 9 -1 0.38 -2 -3 0.38 -7 4 0.25 18 0 0.25 -2 1 0.25 4 IN 7 0.29 25 0 0.29 1 3 0.29 9 Coffee OUT 5 0.29 16 1 0.29 3 9 0.29 31 BAL 3 0.29 9 -1 0.29 -2 -6 0.29 -22 IN 6 0.28 21 0 0.28 0 2 0.28 7 ECI OUT 5 0.28 18 1 0.28 4 6 0.28 21 BAL 1 0.28 4 -1 0.28 -4 -4 0.28 -14 IN 14 0.07 194 4 0.07 56 0 0.07 1 6 0.12 52 2 0.12 13 0 0.12 0 23 0.56 40 8 0.56 14 2 0.56 4 1 0.08 11 0 0.08 0 0 0.08 4 9 0.16 54 4 0.16 24 0 0.16 0 ACV OUT 2 0.07 28 1 0.07 18 1 0.07 18 2 0.12 14 1 0.12 7 1 0.12 10 19 0.56 34 7 0.56 13 14 0.56 25 1 0.08 17 1 0.08 10 1 0.08 10 4 0.16 22 3 0.16 17 3 0.16 18 BAL 12 0.07 167 3 0.07 38 -1 0.07 -17 5 0.12 38 1 0.12 7 -1 0.12 -10 3 0.56 6 0 0.56 1 -12 0.56 -21 0 0.08 -6 -1 0.08 -9 0 0.08 -6 5 0.16 32 1 0.16 7 -3 0.16 -18 IN 25 0.22 116 4 0.22 20 0 0.22 0 83 0.46 180 13 0.46 28 0 0.46 0 Khat OUT 7 0.22 32 2 0.22 10 8 0.22 35 10 0.46 21 3 0.46 7 10 0.46 23 BAL 18 0.22 84 2 0.22 9 -8 0.22 -35 73 0.46 159 10 0.46 21 -10 0.46 -22 IN 10 4 29 22 6 54 19 6 52 23 8 67 17 5 42 Livestock OUT 2 0 0 2 0 0 6 2 1 18 4 3 8 2 2 BAL 8 4 29 20 6 54 13 4 51 5 4 64 10 3 40

Note: Non-bolded nutrient (N/P/K) denotes originally reported macronutrient amount (kg/farm/yr). FS denotes field size adjusted as per land use allocation in respective representative farm. Bolded nutrient (N/P/K) denotes macronutrient amount (kg/ha/yr).

pg. 73

Table 5.7: Farm level macronutrient inflows, outflows and balances (kg/ha/yr) by farm component, across five representative farms.

Farm type Farm type Farm type Farm type Farm type Component Enset-based Enset-coffee Enset-cereal-vegetable Enset-livestock Khat-based N FS N P FS P K FS K N FS N P FS P K FS K N FS N P FS P K FS K N FS N P FS P K FS K N FS N P FS P K FS K IN 0.72 0 0.72 0 0.72 0 0.45 0 0.45 0 0.45 0 0.37 0 0.37 0 0.37 0 0.38 0 0.38 0 0.38 0 0.25 0 0.25 0 0.25 0 Enset OUT 5 0.72 7 1 0.72 1 3 0.72 4 1 0.45 2 0 0.45 0 0 0.45 0 2 0.37 5 0 0.37 0 1 0.37 3 0 0.38 1 0 0.38 0 0 0.38 0 1 0.25 4 0 0.25 0 1 0.25 4 BAL -5 0.72 -7 -1 0.72 -1 -3 0.72 -4 -1 0.45 -2 0 0.45 0 0 0.45 0 -2 0.37 -5 0 0.37 0 -1 0.37 -3 0 0.38 -1 0 0.38 0 0 0.38 0 -1 0.25 -4 0 0.25 0 -1 0.25 -4 IN 0.29 0 0.29 0 0.29 0 Coffee OUT 5 0.29 16 1 0.29 3 9 0.29 31 BAL -5 0.29 -16 -1 0.29 -3 -9 0.29 -31 IN 0.28 0 0.28 0 0.28 0 ECI OUT 3 0.28 10 0 0.28 1 4 0.28 16 BAL -3 0.28 -10 0 0.28 -1 -4 0.28 -16 IN 13 0.07 190 4 0.07 56 0.07 0 6 0.12 52 2 0.12 13 0.12 0 16 0.56 29 8 0.56 14 0.56 0 0.08 0 0.08 0 0.08 0 9 0.16 54 4 0.16 24 0.16 0 ACV OUT 1 0.07 14 1 0.07 14 0 0.07 0 0 0.12 0 0 0.12 0 0 0.12 0 16 0.56 29 5 0.56 9 12 0.56 21 0 0.08 3 0 0.08 1 0 0.08 2 1 0.16 6 0 0.16 0 0 0.16 0 BAL 12 0.07 176 3 0.07 42 0 0.07 0 6 0.12 52 2 0.12 13 0 0.12 0 0 0.56 0 3 0.56 5 -12 0.56 -21 0.08 -3 0.08 -1 0.08 -2 10 0.16 48 4 0.16 24 0 0.16 0 IN 25 0.22 116 4 0.22 20 0.22 0 82 0.46 179 13 0.46 28 0.46 0 Khat OUT 5 0.22 23 2 0.22 9 5 0.22 23 10 0.46 21 3 0.46 7 10 0.46 23 BAL 20 0.22 93 2 0.22 10 -5 0.22 -23 73 0.46 158 10 0.46 21 -10 0.46 -23 IN 0 0 0 5 1 8 0 0 0 1 0 1 3 1 3 Livestock OUT 2 0 0 2 0 0 6 2 1 17 4 3 8 2 2 BAL -1 0 0 3 1 8 -6 -2 -1 -16 -4 -2 -4 -1 1

Note: Non-bolded nutrient (N/P/K) denotes originally reported macronutrient amount (kg/farm/yr). FS denotes field size adjusted as per land use allocation in respective representative farm. Bolded nutrient (N/P/K) denotes macronutrient amount (kg/ha/yr).

pg. 74

6. DISCUSSION

The expansion of khat cultivation has provided a short-term, but unique opportunity to quantify and compare nutrient inflows and outflows of five distinct home garden types under the same conditions. The aim of this research was to produce representative farms for each home garden type, quantify their macronutrient inflows and outflows and compare based on component and farm level nutrient balances to improve understanding of these transitioning systems. In this chapter uncertainties regarding partial nutrient balances and its implications on this research are described in section 6.1. The results of this study are interpreted, discussed and compared to recent literature in section 6.2. In section 6.3 suggestions for improved methodology and possibilities for future research are outlined. To finish, section 6.4 recommends management actions that can be taken to address nutrient deficiencies, and explores the nutrient-related consequences of khat expansion.

6.1 UNCERTAINTIES

Oenema et al. (2003) distinguished possible sources of biases and errors in nutrient balances. In this study, five potential sources of bias were identified: (i) personal biases, (ii) sampling biases, (iii) measurement biases, (iv) data manipulation biases and (v) biases due to fraud.

i. Personal biases. When constructing a nutrient balance, its boundaries are in the opinion of the researcher. The partial nutrient balance was produced for the component and farm level. Parameters (e.g. DEP, BNF, leaching, etc.) have been excluded as quantitative, regional-specific data was not available (Elias et al., 1998). Had outflows such as leaching, denitrification and water erosion been included, the chiefly positive balances may have neared equilibrium or even been pressed into a deficit. However, Elias et al. (1998) determined removal in harvested products (OUT1) and removal in crop residue (OUT3) were the major causes of N and P export from the soil in most fields. This suggests that despite excluding some parameters, OUT1 and OUT3 actually provide a good indication of nutrient removal from the soil. Elias et al. (1998) did point out leaching and denitrification could have a considerable role in nutrient removal, based on estimations and assumptions. However, Elias’ team (1998) questioned the accuracy of this finding because it relied on estimations and assumptions. Due to this and the little conclusive evidence on best practices for leaching, denitrification and water erosion estimations in this specific agro-ecological zone, it was elected to exclude these parameters as their estimation would likely only increase error.

ii. Sampling biases. Within nutrient balance assessments, sampling can be a large potential source of bias when quantifying all nutrient losses, including leaching, volatilization, erosion and runoff (Oenema et al., 2003). Since this study elected to exclude these losses from the research, this potential for bias was more or less excluded too.

pg. 75

iii. Measurement biases. Laboratory analysis of nutrient content was performed for livestock manure and all crop outputs, with the exception of cabbage as it was not sampled by Mellisse et al. (in prep.). Livestock output nutrient content was obtained from literature. The laboratory analysis was completed at Hawassa University College of Agriculture for N content and Wondo Genet College of Forestry and Natural Resources for P and K content. Poor calibration of equipment, incomplete dissolution and rushed extraction of nutrients are all sources of measurement bias (Oenema et al., 2003). Although this research is reliant on laboratory results, the study has little control over these biases but should be mentioned for comprehensiveness.

iv. Data manipulation biases. In this research, inputs and outputs of 63 home gardens were averaged, generalized and grouped by home garden type. Home garden types were developed into representative farm models. These provided the means for comparative analysis but increases data manipulation by simplifying the home garden. As a result, some farms components were never defined, such as faba beans, a legume with potential for BNF but only appeared in 3 of 63 farms. Potatoes, a cash crop, were also never defined as part of a component as it occurred on just 4 of 63 farms. Another example was coffee. The traditional cash crop only qualified for the enset-coffee representative farm, even though it was present in 8 of 18 khat-based farms, narrowly missing the 50% or more qualification cut-off. Data manipulation biases introduced can alter analyzes, but boundaries must be drawn in any nutrient balance assessment.

Laboratory results of nutrient content were also averaged to simplify the quantification of nutrient inflows and outflows. With this you may introduce an inaccuracy. For example, the nutrient content of kocho was averaged from 3 to 7-year-old kocho samples even though N and K content both reduce with age. Another instance was the nutrient content of organic matter (IN2), which was averaged across all home gardens, even though heterogeneity within farms is known to occur.

v. Biases due to fraud. Oenema et al. (2003) refer to biases due to fraud as stakeholders that may manipulate the budget to minimize economic consequences. In this study, it seems exaggerated to accuse farmers of deceitfulness. In lieu of fraud, there can be biases due to farmer error and deliberate or not, farmers can misreport inputs, outputs and land size. When farmers report inputs, some have the tendency to report the recommended dose vs. the actual dose. Whether or not this is a frequent occurrence is difficult to measure, as some farmers may actually apply the recommended application. Recommended doses came about after Murphy (1963) reported survey results demonstrating N and P were limiting crop production in Ethiopia.

Reported inputs may also be skewed if farmers misreport field and farm sizes. To extract nutrient amount, intensification variables are measured in per hectare terms (e.g. kg DAP/ha). Beegle et al. (2012) uncovered under- or overestimation of farm size can amplify measurement errors at smaller farm sizes. A conclusion especially relevant in

pg. 76

this study; where representative farms were 1.10, 1.21, 1.15, 1.12 and 1.03 ha for enset- based, enset-coffee, enset-cereal-vegetable, enset-livestock and khat-based systems, respectively. When the dependant variable is measured in per hectare terms, misreporting farm sizes amplifies per hectare measurement error at very small farm sizes. In this case inputs and outputs are measured per farm, although the average farm size of any home garden is equal to 1.12 ha.

To assess the measurement area, GPS field measurements from 24 farms were plotted against farmer-reported field size. Half over-reported and half under-reported. However when ACV field measurements were excluded from the analysis due to their significant divergence (R2 = 0.1119), 63% of farmers were found to have over-reported their field sizes by an average of 0.2 ha. One-fifth of a hectare on an average farm size of 1.12 ha is substantial. In the quantification of the nutrient balance, it was debated whether to use GPS measured or farmer-reported field sizes. Eventually farmer-reported field sizes were favoured as only 38% of all farms studied had been measured via GPS. In addition, farmer-reported values were recorded six months prior to when the GPS measurements were taken. As such annual components, like annual cereals and vegetables had already been harvested and replaced with another crop or laid fallow, rendering a comparison of GPS measured versus farmer-reported values at odds with one another.

Errors can originate from spatial and temporal variability and show up as variance in results. Two error types were also identified: (i) sampling errors and (ii) measurement errors.

i. Sampling errors. The nutrient balance assessment in this study is a ‘snapshot in time’ as it considers all inflows and outflows over one year. Therefore, it gives an indication of the nutrient balances within that time span, but can say little about balances over temporal scales or extrapolate across spatial scales. When sampling organic matter (IN2) and crop outputs, there is always variance in soils, crops and animal wastes. This is even the case when balance margins are strict, as is the case with this research.

ii. Measurement errors. Variations introduced in the determination of volume and composition of samples can result in measurement errors. The study is unlike from similar, research (Abrham, 2014; Elias et al., 1998) as it analyzed the nutrient content of all crop outputs (OUT1) and organic matter (IN2). It relies heavily on the accurate measurement of this content for its analysis.

6.2 INTERPRETATION AND DISCUSSION OF RESULTS

The results are interpreted by discussing the nutrient inputs, outputs and balances initially on the farm-scale and then narrowing into the component level, by farm component. To assist in comparative analysis, the results originally reported in kg/farm/yr have been converted to kg/ha/yr (Table 5.7). The livestock component has been excluded from this conversion as it is

pg. 77

not possible to express livestock inflows and outflows per hectare. In this discussion, the expressions ‘very strong’, ‘strong’, ‘moderate’ and ‘slight’ are used to describe nutrient balances. The classification was originally put forward by Smaling (1993) and used in the nutrient balance assessment by Elias et al. (1998) to classify depletion. As such, the original terms refer to nutrients lost, but this discussion will apply the same ranges for nutrients accumulated (Table 6.1). Since Elias et al. (1998) only classified N and P; K classification will be based on half of the suggested N ranges, as K balances were generally found to be half of N balances in this nutrient balance assessment.

Table 6.1 Nutrient balance analysis interpretation criteria (expressed as kg of nutrient lost (or added)/ha/yr.

N P K Classification + - + - + - Very strong > 40 < - 40 > 7 < - 7 > 20 < - 20 Strong 20 to 40 - 20 to - 40 4 to 7 - 4 to - 7 10 to 20 - 10 to - 20 Moderate 10 to 20 - 10 to - 20 2 to 4 - 2 to - 4 5 to 10 - 5 to - 10 Slight < 10 > - 10 < 2 > - 2 < 5 > - 5

6.2.1 FARM SIZE

Agricultural economists have expressed concern over farmers’ self-reporting of land size (Carletto et al., 2013). To validate this critique, Mellisse et al. (in prep.) measured field and farm size of 24 of the 63 surveyed farms using Global Positioning System (GPS) devices. The GPS measurements put side by side with farmer-reported dimensions Figure 5.2. The divergence of ACV is attributed to the time lag between farmer reporting and GPS measurement. Mellisse et al. (in prep.) reported this was done six months apart at which point farmers have either harvested their ACV or switched to another crop. This also explained the coefficient of determination (R2) value of 0.1119 (Table 5.1) for ACV. For coffee, ECI, and khat were R2 values of 0.6904, 0.5700 and 0.5737, respectively. A possible explanation could be the nature of these crops as cash crops. Farmers may be more likely to under-report their land size of profitable crops, especially when requested to report to government officials.

6.2.2 FARM LEVEL NUTRIENT BALANCES

Table 6.2 presents only the farm nutrient balances by representative farm. At the farm level, across all five representative farms, N balances were moderately to very strongly positive. P also had mainly positive balances, but fluctuated from slight to very strong balances. Two representative farms had moderate to strong negative P balances which was the enset-coffee farm system and enset-cereal-vegetable system, respectively. K had moderately negative balances across all representative farms; except for enset-based which was only slightly negative. The results at farm level are logical. The mineral fertilizer urea provides high N input

pg. 78

(45 kg N/100 kg urea), DAP also provides N input (18 kg N/100 kg DAP) and a small quantity of P (20 kg P/100 kg DAP). There is no input source of K at the farm level. These farm balances are a sharp contrast to Stoorvogel, Smaling and Janssen’s (1993) national report on Ethiopia’s nutrient depletion to be rising at -122 kg N/ha/yr, -13 kg P/ha/yr and -41 kg K/ha/yr, which were very strong depletions according to Smaling’s (1993) own interpretation criteria. However, extensive landscape differences make direct comparisons of a nationwide balance to a woreda (district) specific balance—such as this study—nearly impossible. Moreover, the Stoorvogel, Smaling and Janssen assessment took place over twenty years ago. In this period, inorganic fertilizer use rose dramatically (Abate et al., 2015; Kiros et al., 2012; Wallace & Knausenberger, 1997) and could at least partially explain why Ethiopia’s soil was so depleted in 1993. In fact, Stoorvogel, Smaling and Janssen (1993) found the most depleted nutrient to be N (-122 kg N/ha/yr), the nutrient most prevalent in mineral fertilizers.

Ten years later, Roy et al. (2003) also estimated Ethiopia’s nutrient balance to be negative for all macronutrients, with losses of -47, -7 and -32 kg/ha/yr of N, P and K, respectively. These deficiencies were considered very strong (Table 6.1; Smaling, 1993). Compared to Stoorvogel, Smaling and Janssen (1993), Ethiopia’s nutrient levels had improved. Since, N and P balances had especially improved and were less negative; this could be evidence of increased mineral fertilizer use across Ethiopia. Again, comparing this study’s assessment to those on a national scale should be preceded with caution. Ethiopia’s landscape, soils and crop cultivation are highly diverse. However, since this research had several cases of positive nutrient balances, the popular suggestion that all Ethiopian soil suffers from nutrient mining, certainly cannot be generalized across the country.

Table 6.2 Farm partial nutrient balances (kg/ha/yr) by representative farm. The livestock component is excluded, but the internal input of compost is included. Representative farm N P K Enset-based 169 40 -4 Enset-coffee 12 -2 -9 Enset-cereal-vegetable 20 -6 -5 Enset-livestock 20 2 -7 Khat-based 76 10 -7

Negative nutrient balances are often granted as evidence of soil nutrient depletion on the farm, national or larger scale. Frequently, they are a call for alarm. Ethiopian smallholder farms are sources of survival and assumed to be continuously farmed, with limited time to lie fallow and have nutrients restock. As a result, it has been widely acknowledged these plots and farms must have nutrient depleted soils. However, Vanlauwe and Giller (2006) argue not all nutrient balances are always negative. In fact, some plots have very high positive balances, likely through concentration of nutrients from other parts of the farm (Vanlauwe & Giller, 2006).

Table 6.2 shows an exceptionally strong N balance for the enset-based farm, compared to all other representative farms. This is a surprising finding as enset-based farms receive relatively

pg. 79

little external input. Upon further investigation, the N surplus was deemed particularly high due to its conversion to kg/ha/yr. The enset-based farm cultivates some ACV on a very small plot of land (0.07 ha) and when its original mineral fertilizer input (IN1) of 12 kg N/farm/yr was converted to hectare basis, the value increased dramatically.

The nutrient balance assessment by Elias et al. (1998) was the most spatially relevant to this research as they assessed farming systems in the Kindo Koisha district of Wollaita in southern Ethiopia, some 150 km west of Sidama and Gedeo. Comparing on this regional scale, Elias et al. (1998) found N balances to be negative across all household groups (rich, medium, poor and very poor) and the P balance to be positive for most farms (Table 6.3). However, their study estimated removal in leaching, denitrification and water erosion. The result was an uncertainty range thought to include the ‘real’ value (Roy et al., 2003). The inclusion of these parameters could explain the shift to a negative N farm balance in the Elias et al. (1998) assessment versus the positive farm balance observed in this research.

Table 6.3 Farm nutrient balances (kg/ha/yr) for different household groups (Elias et al., 1998; adapted from Roy et al., 2013).

Households N P

Rich -47 11.7 Medium -51 4.8 Highland Poor -19 3.6 Very poor -6 1.1 Rich -49 30.5 Medium -41 17.3 Lowland Poor -55 3.8 Very poor -20 -1.6

Elias et al. (1998) did not quantify the K balance as they identified only N and P as particularly deficient. In the paper, Elias et al. argued that potassium was commonly available in Ethiopian soils and sufficient enough to satisfy crop requirements. Yet, K had moderately negative balances in four representative farms, a slight negative balance in one representative farm and the only nutrient to have negative balances across all representative farms. Based on this finding, one may conclude K is in fact the most important nutrient to include in a balance assessment. The farm level analysis does not present a full representation of what occurs within a home garden system. The following analysis will be on the component level.

6.2.3 COMPONENT LEVEL: ENSET

N and K enset nutrient balances reveal very strong surpluses in the enset-based system (Table 6.4). These fields had accumulations of 117 kg N/ha/yr and 34 kg K/ha/yr, whereas the second highest N surplus was 33 kg N/ha/yr (enset-cereal-vegetable system) and the second highest P

pg. 80

surplus was 6 kg P/ha/yr in the same system. This finding was notable as enset-based farms have the smallest allocation of grazing land (6%) and smallest average TLU (0.43) (Figure 5.3). One possible explanation is the practice of poorer, enset-based farmers allowing richer farmer’s cattle graze on their land and in return collect the livestock’s manure for their own use. This could be 5-6 cows at any given time. Enset-based farmers may lack large grazing land plots but they have a large supply of internal fodder, especially enset leaves, which could potentially make up the difference.

Table 6.4 Enset component nutrient balances (kg/ha/yr) by home garden type.

Representative farm Component N P K Enset-based Enset 117 1 34 Enset-coffee Enset 5 -1 -1 Enset-cereal-vegetable Enset 33 -2 6 Enset-livestock Enset 9 -2 -7 Khat-based Enset 18 -2 4

Enset-coffee farms also had small grazing land (7%) and low average TLU (0.46) yet lacked the high N surpluses found in enset-based systems. Enset-coffee farmers may practice the same livestock grazing for manure trade, but may spread out their acquired organic matter across their enset, coffee and enset-coffee intercropping fields. These plots receive exclusively organic matter as their sole input.

6.2.4 COMPONENT LEVEL: COFFEE AND COFFEE + ENSET INTERCROPPING

The traditional cash crop only appeared in enset-coffee systems. Its counterpart, intercropping of enset and coffee was also only present in this representative farm. Despite its organic matter inputs, the P balances were slightly and moderately negative and the K balances were strong and very strongly negative, suggesting organic matter alone, at least in its current form, may not be sufficient for these components. The N balance was slightly positive with 9 N kg/ha/yr and 4 N kg/ha/yr on coffee and ECI plots, respectively (Table 6.5).

Table 6.5 Coffee and enset + coffee intercropping (ECI) component nutrient balances (kg/ha/yr) by home garden type.

Representative farm Component N P K Coffee 9 -2 -22 Enset-coffee ECI 4 -4 -14

Negative P and K balances for coffee simply promote arguments in favour of khat cultivation versus the age-old practice of coffee farming. If this persists, coffee cultivation may deplete coffee and ECI plots of its P and K. However, temporal projections should be treated with caution in nutrient balance assessments. Enset-coffee farms are the lowest receivers of mineral

pg. 81

fertilizers likely due to their small ACV allocation (10% land share) compared to enset-based farms with the second largest ACV allocation (28% land share). Recommendations to boast P & K in coffee and ECI plots are put forward.

6.2.5 COMPONENT LEVEL: ANNUAL CEREALS AND VEGETABLES

Annual cereals and vegetables were present in every representative farm and revealed mainly slight to very strong positive N balances, mainly slight to very strong positive P balances and moderate to very strong negative K balances (Table 6.6). N also had one slight negative balance and P had one very strong negative balance. These findings indicate nutrient balances for ACV vary substantially. ACV is the only component to receive mineral fertilizer (IN1) with sporadic quantities of organic matter (IN2), which could explain its negative K balances. The extremely ‘very strong’ N and P balances in the enset-based system are due to the system’s small land allocation for ACV (0.07 ha).

Table 6.6 Annual cereal and vegetable (ACV) component nutrient balances (kg/ha/yr) by home garden type.

Representative farm Component N P K Enset-based ACV 167 38 -17 Enset-coffee ACV 38 7 -10 Enset-cereal-vegetable ACV 6 1 -21 Enset-livestock ACV -6 -9 -6 Khat-based ACV 32 7 -18

In the Kindo Koisha district, Elias et al. (1998) also analyzed the nutrient composition of annual cereals and vegetables. However, the only overlapping crop with this study was maize. The report only considered N and P, but they found the nutrient composition (%DM) for maize was 1.25 and 0.18, respectively. This study found 1.13 and 0.54 (%DM) for maize, a comparable nutrient composition. All values fit into the Stoorvogel and Smaling (1990) reported mean values from several countries across sub-Saharan Africa, except for 0.54 %DM of P which exceeded Stoorvogel and Smaling’s (1990) 0.15-0.27 range. However, these are the lowest and highest quartiles from several countries and do not represent regional variation.

6.2.6 COMPONENT LEVEL: KHAT

Khat, unsurprisingly, received the highest quantity of mineral fertilizers, with an extremely ‘very strong’ positive N balance of 159 kg N/ha/yr in a khat-based farm. Traditionally, only enset and coffee receive organic matter and mineral fertilizers are reserved for khat. Occasionally organic matter is applied to khat plots, but only in the khat-based representative farm. This is likely because there was an excess of organic matter and farmers want to encourage profitable khat cultivation any way possible. Despite occasional compost application, khat fields often have the

pg. 82

most severe K deficiencies (Table 6.7). However, before farmer’s flock to compost piles to supply khat plots with increased K, other farm component nutrient balances should be considered. Precaution should also be taken with neutral balances. This indicates the soil fertility was narrowly maintained for this cropping season. In such close situations it is advisable to supply more nutrients (Abrham, 2014).

Table 6.7 Khat component nutrient balances (kg/ha/yr) by home garden type.

Representative farm Component N P K Enset-livestock Khat 84 9 -35 Khat-based Khat 159 21 -22

At present there are no recommended doses available to farmers for khat cultivation. As a consequence, farmers rely on the blanket fertilizer recommendations for annual cereals and vegetables. Evidently, this method is supplying khat fields with very high N and P application rates leading to a very strong positive N balance, very strong positive P balance and very strong deficiencies of K. Urea and DAP reductions and compost (combined with animal manure) increases could address this. In other words, khat plots require integrated nutrient management of mineral and organic fertilizers.

At the time of data collection (2014/15) global khat markets still existed. It was the last cropping season, before the Netherlands and the United Kingdom (UK), Europe’s last legal khat nations, banned khat. The UK was Ethiopia’s third largest khat export destination, just behind Djibouti (2nd) and Somalia (1st). It is said the UK was a key hub for smuggling khat to the United States (US). Political instability in Yemen, another popular khat export destination, has closed the airports and hindered imports. At present, only domestic and two regional (Somalia, Djibouti) markets for khat remain.

Khat export earnings had been steadily rising prior to the bans, exporting 36 000, 41 000 and 41 100 tonnes in 2010, 2012, 2013, respectively (Fantahun, 2015). Ethiopia brought in 209, 238 and 297 million US dollars during those years. Despite its continued rise in domestic production, the khat export declined 8.4% in the 2014/15 fiscal year.

While this study cannot make market projections or temporal nutrient predictions, the ban will undoubtedly have an impact on the demand for khat. If the demand stays the same, it either indicates domestic demand has increased (with troubling health concerns) or an even larger black market to export the drug has emerged. Khat is addictive and can have devastating impacts on labour productivity. For instance, in Yemen where 90% of men are estimated to chew khat up to six hours a day, labour productivity in peak hours is low (WHO Bulletin, 2008). The Ethiopian Ministry of Agriculture have confirmed the cultivation and distribution of khat is operated solely by the farmers, with no support from authorities. Should the government remain indifferent to khat farming, farmers may seek other market opportunities and perhaps even returning to coffee production for their cash crop income.

pg. 83

6.2.7 COMPONENT LEVEL: LIVESTOCK

Ethiopia’s livestock population is said to be the largest in Africa (CSA, 2009). Ethiopians assign high personal wealth and cultural value to the quality and quantity of their livestock. The replacement of enset with khat monoculture has induced internal fodder shortages. Mellisse et al. (in prep.) found this to have direct repercussions on per capita herd size, herd composition and the nutritional value of household diets. The main animal products sold off-farm are sourced from these livestock. More than any other component, livestock had slight to moderate positive balances for N, moderate to strong positive balances for P and very strong positive balances for K (Table 6.8). The following balances are presented in kg/farm/yr as the livestock component (not a crop) cannot be converted to kg/ha/yr.

Table 6.8 Livestock component nutrient balances (kg/farm/yr) by home garden type.

Representative farm Component N P K Enset-based Livestock 8 4 29 Enset-coffee Livestock 20 6 54 Enset-cereal-vegetable Livestock 13 4 51 Enset-livestock Livestock 5 4 64 Khat-based Livestock 10 3 40

The livestock balance analyzed two inputs (IN3; IN4) and two outputs (OUT3; OUT4). The livestock balance was likely skewed positively as it does not account for livestock that remains and circulates within the system for years. For this reason, livestock systems are inherently different from cropping systems and its outputs from animals and animals products sold off- farm (OUT3) and household consumption (OUT4) are only one part of the equation. In addition, two other outputs were excluded: 1) household consumption of milk and eggs was excluded because there was no explicit data collected, and 2) manure as output from livestock was excluded because there were no composite samples taken of fresh manure. Fresh manure and compost differ as compost is the mixture of manure and household refuse and through collection and storage has lost some of its original nutrient content.

The K influx from internal fodder (IN3) is 67 kg K/farm/yr at its largest in an enset-livestock system. The K content of enset leaves is large (4.60%) and meant high K input. The K surplus is revealed in Table 6.8 across all representative farms, but especially in the enset-livestock system (64 kg K/ha/yr). Enset leaves supplement grasses collected from enset, coffee and khat fields. Relative to nutrient inflows from internal fodder, nutrient inputs from external fodder (IN4) was small. The largest input came in an enset-coffee system with 14 kg NPK/farm/yr. Even the enset-livestock supplied only 3 kg NPK/farm/yr of external fodder to its livestock. Enset- coffee farmers spend an average of 180 ETB ($8 USD) on external fodder. This represents 7% of their average total input costs. For such small nutrient inflows, this money may be better spent on other farm needs.

pg. 84

Ammonium volatilization of manure is likely a large nutrient loss factor in home gardens, particularly in the livestock component. In this study, only the nutrient composition of compost (including manure) was analyzed. Ammonium losses are amplified when manure is repeatedly handled, stockpiled whilst moist and/or not used immediately. This is especially true in warm, dry conditions similar to those in the study area. Field observations revealed the organic matter was stored in an uncovered pile close to the homestead. Nutrient composition of fresh manure was not tested so average losses cannot be estimated. Nor could average manure production based on TLU derive the amounts of manure necessary to meet nutrient requirements. However, Davies et al. (2009) estimated N losses vary from 5-50%, P losses vary 3-30% and K losses vary from 5-80%. Considering these potential losses, integration of better manure handling techniques has vast potential to retain higher nutrient composition of manure, possibly correcting nutrient deficiencies. Overall, when negative nutrient balances are observed, they are marginal, with the largest at -9 K kg/ha/yr.

This study took composite samples of compost, including manure, not fresh manure. Compost (IN2) is applied to farmer’s fields, while fresh manure is collected, mixed with household waste and stored in an outdoor pile until applied. Similar composition data of this regionally-specific sort of compost could not be located in literature, but Elias et al. (1998) did sample fresh manure on similar home gardens. Elias et al. studied farms in the highlands and lowlands and sampled manure from each. For the purposes of this comparison, only the highland values are presented as this study’s farms were in the highlands and midlands. Elias et al. (1998) also provided a range collected from literature data. A range for compost found for Central Kenyan farms (Lekasi et al., 2003; Kimani & Lekasi, 2004, as cited in Paul et al., 2009) is also included for a comparison of compost (Table 6.9).

Table 6.9 Nutrient composition (%) (in dry matter) of manure (Elias et al., 1998), Central Kenyan compost (Lekasi et al., 2003; Kimani & Lekasi, 2004) and compost (this study).

Material N (%) P (%) K (%) Manure (Elias et al., 1998) 1.68 0.23 Not sampled Manure from literature data (Elias et al., 1998) 1.1 – 1.7 0.13 – 0.26 Not included Central Kenyan compost (Lekasi et al., 2003; 1.12 (0.3-1.9) 0.3 (0.1-0.8) 2.4 (0.4-7) Kimani & Lekasi, 2004) Compost (IN2) 0.83 0.03 0.29

Compared to the manure nutrient content from Elias et al. (1998), the compost from this study has twice as less N and seven times less P. This study’s compost nutrient levels are lower than the lowest end of the range provided from literature data (Elias et al., 1998). However, Elias and colleagues (1998) provided values for manure which would be expected to be greater than that of compost, regardless. In a more direct comparison to Central Kenyan compost, this study’s compost does not fare any better. In fact, the P content given for Central Kenyan compost (0.3% P) exceeds that of manure by Elias et al. (1998) (0.23% P). N content of this study’s compost (0.83% N) was the only nutrient to fall into the Central Kenyan compost range

pg. 85

(0.3%-1.9%). Overall, when compared to nutrient values of manure and compost from literature, this study’s compost is low in all macronutrients.

6.3 METHODOLOGICAL IMPROVEMENTS AND SUGGESTIONS FOR FURTHER RESEARCH

During the course of this research, three topics that were beyond the scope of this research were distinguished. These topics relate to parameter exclusion, comparative analysis and assembly of the ‘ideal’ home garden. Data availability prevented a ‘complete’ nutrient balance assessment. Quite often balance assessments rely on assumptions for their inflows and outflows. In lieu of assumptions, this research elected to use farmer-reported survey data, laboratory analysis of composite samples, field observations and literature data, but only when a crop was not sampled. As a result of not assuming the remaining processes not covered under this data collection, not all nutrient losses were accounted for.

Many nutrient balance assessments use the earlier developed methodology by Stoorvogel and Smaling (1990) but use different transfer functions to estimate deposition, sedimentation, leaching and erosion. This results in methodological discrepancies and hinders regional, national and continental comparison. Lesschen et al. (2007) have aimed to improve the existing methodology by making it spatially explicit. Their study upgrades transfer functions and explicitly models the uncertainties in estimations. Further research should demonstrate the effectiveness of this improved methodology.

For future balance assessments, composite samples should be taken for fresh manure. This would add a significant outflow to the livestock component and give a more accurate representation of the internal nutrient processes. Also a database of regionally specific assumptions for commonly excluded parameters could be constructed using the improved methodology (by Lesschen et al., 2007). This could also improve comparative analysis between nutrient balance assessments.

The ‘partial’ nutrient balance assessment as completed in this research really is partial. It considers nutrient inputs, removals and recycling, but excludes nutrient stocks. To improve the balances, more soil- and field-level measurements are necessary. Good sampling strategies are crucial as soil properties are highly variable. Roy et al. (2003) insists soil property indicators such as: clay/silt/sand content, pH, organic carbon, etc., need to be readily measurable in order to permit the examination of the actual impact of alternative farming strategies. Moreover, if the research aim is to influence management policies, nutrient balances are a much better indicator of soil fertility status if original nutrient stocks are taken into account.

This study compared farm level balances and component level balances to one another. From these comparisons, it could be tempting to assemble the ‘ideal’ home garden by combining a model of mixed farm components. However, this depends on several factors beyond the scope

pg. 86

of this study, such as the socioeconomic status and personal wishes of the farmer. Socioeconomic status plays a role on the farmer’s accessibility to external inputs (Mellisse et al., in prep.). Further research could possibly establish the most ‘sustainable’ home garden but sustainability would require clear definition to accomplish this. With a model home garden in place, and its effectiveness confirmed, it could be suggested to farmers as an alternative for khat monoculture.

6.4 MANAGEMENT RECOMMENDATIONS

Complete management decisions would require a complete nutrient balance assessment. However, the partial nutrient balance assessment can provide a tool for management purposes. Through the use of nutrient hotspot identification, three management recommendations are suggested. These measurements relate to enset leaves as crop residue or compost additive (6.4.1) and proper manure handling (6.4.2). To finish, the implications of khat expansion on nutrient flows of the home garden are summarized (6.4.3).

Identifying nutrient hotspots indicate either losses or accumulation of nutrients. Slight to very strong negative balances (losses) should be addressed to prevent further depletion. Moderate to very strong positive balances (accumulation) should be addressed to exploit underused nutrients in other areas of the farm. Slight positive balances should remain to provide a nutrient ‘buffer’ for future cropping seasons.

6.4.1 ENSET LEAVES AS CROP RESIDUE OR COMPOST ADDITIVE

On the farm level, K is moderately deficient for each representative farm. On the component level, slight to very strong K deficiencies are virtually ubiquitous across all components amongst all representative farms, with a small exception of positive balances of enset components on three representative farms. The other exception to the rule is the very strong positive K balances of livestock components (Table 6.8). These very strong K balances should be exploited to address the widespread K deficiencies. The positive K balances are attributed to the large K content (4.60%) of enset leaves, the primary ingredient in internal fodder. Although composite samples on fresh manure were not collected, we can derive from the low nutrient composition of compost that its K content (0.29%) may not be as high as it could be. Instead of feeding all enset leaves, or selling excess for small profit, the leaves should be chopped and directly applied to enset, coffee, ECI, ACV and khat fields. Presently, all naturally-occurring grasses on coffee, ACV and khat fields are harvested for additional livestock feed. Grasses have some K (1.96%) but enset leaves boast over twice the K content. Adding enset leaves as a direct crop residue could not only address K deficiencies but also replenish soil organic matter and improve soil physical properties. This is especially relevant on ACV and khat fields which currently receive no or very little application of organic matter.

pg. 87

Another option is indirect application by adding chopped enset leaves to compost. Enset leaves fed to livestock will eventually return to the field as animal waste, but will likely undergo large nutrient losses before then. Utilizing enset leaves as a compost additive will supplement some of this nutrient loss, especially K.

6.4.2 PROPER MANURE HANDLING

As seen in the comparative analysis of this study’s compost with that in the literature, this study’s compost is low in N, P and K. This suggests compost is subject to large nutrient losses, which result in lower-nutrient compost. Under manure handling, losses are by far highest for N, followed by K and very minimal for P. At present, compost is stored in an outdoor, uncovered pile, also referred to as farm yard manure in the literature (Davies et al., 2009). In Zimbabwe, better manure quality and higher maize yields deduced losses of ammonia N were lower for manure composted anaerobically (in a pit) than composted aerobically (in an open-air heap on the ground) (Davies et al., 2009). The risk for N losses are said to increase with more aerobic storage systems (above-ground heaps). Rotz (2004) also suggested undisturbed natural ‘crusts’ on top of a manure pit may substantially reduce ammonia losses. In the local context, whole enset leaves placed over top of a dug-out pit of compost, may help retain better nutrient content.

K losses, unlike N losses, are not associated with high temperatures, but rather its ability to retain the liquid portion of the manure. The most common K loss from compost is from the leaching of soluble nutrients, particularly from urine. Urine is inherently difficult to manage, but remains the most important process to harness K content in compost. Urine is ideally stored in a closed pit. Another option is that farmers can add urine to manure heaps, but this method would require a water-tight base to collect leached liquids. Neither of these options is practical in the local context as these improved systems demand higher labour and financial investment in storage facilities. Cheap and available ‘makeshift’ water-tight bases such as plastic or aluminum sheets could be tested, but further experimentation would be necessary to test effectiveness. Farmers could also collect urine separately and drain the urine to perennial crops via channels. However, even the collection of urine poses a problem as livestock is confined to grazing land and their urine will infiltrate almost immediately. An additional suggestion to reduce K loss would be to apply manure directly as a nutrient source. Still this comes with concerns about the quality of manure, as composting allows the killing of weed seeds, elimination of pathogens and reduction of odor problems.

Losses of P may diminish under these manure handling strategies, but original P content in manure tends to be very low, with losses already minimal. However, there is some potential for the improvement of compost management, but it will mainly improve N losses. Implementing the suggested manure handling practice of storing compacted compost in an enset leaf-covered pit, will reduce N loss and possibly lessen K losses.

pg. 88

6.4.3 NUTRIENT-RELATED CONSEQUENCES OF KHAT EXPANSION

Recommendations to integrate organic matter and mineral fertilizer on nutrient depleted fields (e.g. coffee, ACV and khat) were not offered because increasing organic matter is an unlikely option, especially in light of khat expansion. Mellisse et al. (in prep.) have found the replacement of enset with khat monoculture to induce internal fodder shortages, particularly enset leaves. This was established to have direct repercussions on per capita herd size and herd composition. Abebe (2013) reported when livestock holding is low, manure production decreases and results in reduced enset yields due to lack of organic matter. Instead, the management recommendations put the focus on maximizing the nutrient accumulation revealed in the component and farm level nutrient balance assessments. However, they rely heavily on the role of enset. Abebe (2013) argues, “as enset produces the highest volume of food per unit area and time, and due to its different end uses and diverse ecological roles, the future of these home gardens depend on the maintenance of enset-based staple food production,” (p. 36). The diversity and integration of these home gardens uphold their stability and resilience. Expansion of khat monoculture not only threatens very strong K deficiencies, but forces home gardens into specialization. Ultimately putting well-established internal nutrient flows in jeopardy. Khat’s profitability may prove to replace coffee as the primary cash crop in Sidama and Gedeo, but tremendous caution should precede khat expansion to the detriment of enset cultivation. Therefore, strategies should be developed to rapidly reverse khat development at the expense of enset.

pg. 89

7. CONCLUSIONS

The aim of this research was to improve the understanding of inflows, outflows and internal flows that make up the nutrient balance of transitioning home garden systems. In recent decades, population pressure induced land fragmentation has driven home gardens to rapidly replace enset and coffee with khat. For proper insight in to the dynamics of this transition, representative farms were conceived for each home garden type—enset-based, enset-coffee, enset-cereal-vegetable, enset-livestock and khat-based—to illustrate which farm components were most significant in each type. The representative farms revealed enset, ACV and grazing land were prevalent across all types, coffee only existed in the enset-coffee type and khat was prevalent in the enset-livestock and khat-based type. Macronutrient (NPK) inflows, outflows and internal flows were quantified and the resulting balances were compared at the component and farm level.

At the farm level, N balances were moderately to very strongly positive, P balances were strongly negative to very strongly positive and K balances were moderately negative. On the component level, N balances were slightly positive to very strongly positive and P balances were very strongly negative to very strongly positive, across all representative farms. K balances were moderately to very strongly negative, with the exception of the enset and livestock components. P balances fluctuate considerably based on internal and external inputs to components and its narrow interpretation criteria.

Some inherent flaws to nutrient balance assessments and this study’s methodology were outlined. Methodological improvements and possibilities for future research included a database of regionally specific assumptions to aid local scale comparative analysis and soil nutrient stock analysis to better indicate soil fertility status.

Most significantly, the balances revealed nutrient hotspots of very strong K depletion in the khat component and a hotspot of very strong K accumulation in the livestock component. To capitalize on the underutilized supply of K, a suggestion to use enset leaves as crop residue or as a compost additive was offered. Use as a crop residue could directly boast the soil’s K content. Review of literature showed the compost in this study has low macronutrient content in contrast to that in similar systems. Recommendations for proper manure handling were given. Although potential for reductions from manure nutrient loss exist, proper handling will likely improve N and only partially improve the K content.

To conclude, it proved valuable to develop not only a farm level nutrient balance assessment, but also a component level assessment, as it revealed the inherent diversity and complexity of home garden systems. Well-established internal nutrient flows sustain home gardens and component level analysis allowed comparison between these flows. Khat expansion threatens internal flows in a positive feedback. When khat expansion induces internal fodder shortages, it causes livestock holding to decrease, which cuts manure production. As a result, there is

pg. 90

reduced enset cultivation, smaller enset yields and a consistent decline of internal fodder, especially enset leaves.

Under current trends, khat will likely replace coffee as the principal cash crop in the region’s home gardens (Mellisse et al., in prep.). This will intensify nutrient mining and induce further K deficiencies, a shortage which could be effectively addressed with the management recommendations put forward. However, these proposals are dependent on adequate enset leaf supply. If khat expansion reduces leaf supply, strategies should be urgently developed to reverse khat development at the expense of enset plots. For centuries, the long-term stability and sustainability of these home gardens were attributed to its intimate association with enset cultivation. The resilient enset plant has been hailed as the ‘tree against hunger’ (Spring et al., 1997) and contributed to the environment by improving nutrient balances in soils (Elias et al., 1998). Now, in the face of khat expansion, enset leaves may just provide the means to secure the survival of the home garden system.

pg. 91

8. REFERENCES

Abate, T., Shiferaw, B., Menkir, A., Wegary, D., Kebede, Y., Tesfaye, K., Kassie, M., Bogale, G., Tadesse, B. & Keno, T. (2015). Factors that transformed maize productivity in Ethiopia. Food Security 7: 965–981.

Abebe, T. (2005). Diversity in home garden agroforestry systems of southern Ethiopia. PhD Dissertation, Wageningen University, The Netherlands: 1-143.

Abebe, T., Wiersum, K. F., & Bongers, F. (2010). Spatial and temporal variation in crop diversity in agroforestry homegardens of southern Ethiopia. Agroforestry Systems 78(3): 309-322.

Abebe, T. (2013). Determinants of crop diversity and composition in Enset-coffee agroforestry homegardens of Southern Ethiopia. Journal of Agriculture and Rural Development in the Tropics and Subtropics 114(1): 29–38.

Abera, K. (2013). Growth, productivity and nitrogen use efficiency of maize (Zea mays L.) as influenced by rate and time of nitrogen fertilizer application in Haramaya District. Eastern Ethiopia.

Abrham, B. (2014). Estimating Soil Nutrient Balance Of Cereal Lands Of Tigray Region, Northern Ethiopia (Doctoral dissertation, AAU).

Agricultural Research Council [ARC]. (1984). The Nutrient Requirements of Ruminant Livestock in: Commonwealth Agricultural Bureaux, Farnham Royal, UK.

Amede, T. & Taboge, E. (2007). Optimizing Soil Fertility Gradients in the Enset (Ensete ventricosum). Chapter 26: Systems of the Ethiopian Highlands: Trade-offs and Local Innovations. In Advances in Integrated Soil Fertility management in Sub-Saharan Africa: Challenges and Opportunities, Springer Netherlands: 289-297.

Araya, H., & Edwards, S. (2006). The Tigray experience: A success story in sustainable agriculture. Environment and Development Series 4, Third World Network, Penang. Retrieved September 8, 2015 from http://www.twnside.org.sg/title/end/ed04.htm.

Beegle, K., Carletto, C., & Himelein, K. (2012). Reliability of recall in agricultural data. Journal of Development Economics 98(1), 34-41.

Blazy, J. M., Ozier-Lafontaine, H., Doré, T., Thomas, A., & Wery, J. (2009). A methodological framework that accounts for farm diversity in the prototyping of crop management systems. Application to banana-based systems in Guadeloupe. Agricultural Systems 101(1), 30-41.

pg. 92

Brandt, S., Spring, A., Hiebsch, C. J., McCabe, T., Tabogie, E., Diro, M., Wolde-Michael, G., Yntiso, G., Shigeta, M., & Tesfaye, S. (1997). The 'Tree Against Hunger' Enset-Based Agricultural systems in Ethiopia, American Association for the Advancement of Science.

Carletto, C., Savastano, S., & Zezza, A. (2013). Fact or artifact: The impact of measurement errors on the farm size–productivity relationship. Journal of Development Economics 103: 254– 261.

Central Statistical Agency of Ethiopia [CSA]. (2007). The 2007 Population and Housing Census: Results of Southern Nations Nationalities Peoples Region. Retrieved September 3, 2015 from http://unstats.un.org/unsd/censuskb20/Attachment489.aspx.

Cherry, K. A., M. Shepherd, P.J.A. Withers, and S. J. Mooney. 2008. Assessing the effectiveness of actions to mitigate nutrient loss from agriculture: A review of methods. The Science of the Total Environment 406: 1–23.

CSA. (2007, December). Livestock Sample Survey (AgLVS 2006), version 1.0. Retrieved September 6, 2015 from http://213.55.92.105/nada4/index.php/catalog/238.

Cuttle, S.P. (2002). Nutrient balances as a tool for researchers and farmers. In: Powell, Jane and et al., (Eds.) Proceedings of the UK Organic Research 2002 Conference, Organic Centre Wales, Institute of Rural Studies, University of Wales Aberystwyth: 153-156.

Davies, O., Wouters, A. P., Gachimbi, L., Zake, J., Ergano, K., Abduke, M., & Keulen, H. V. (2009). Cattle manure management in East Africa: Review of manure quality and nutrient losses and scenarios for cattle and manure management (No. 258, p. 47). Wageningen, the Netherlands: Wageningen UR Livestock Research.

Dessalegne, L., Yesuf, M., Bekele, A., & Abera, D. (2012). Home gardens for food and nutrition security: National appraisal and a case study in Meki-Zway area. Study on Home Garden Vegetable and Fruits Production in Ethiopia (CASCAPE): 1-73.

Dobermann, A.R. (2005). Nitrogen use efficiency–state of the art. University of Nebraska– Lincoln Agronomy & Horticulture Faculty Publications: 316-332.

Elias, E. (2000). Soil enrichment and depletion in southern Ethiopia. International Institute for Environment and Development: 62-82.

Elias, E. (2002). Farmers’ perceptions of soil fertility change and management. ISD and SOS- Sahel International (UK). EDM Printing Press. Addis Ababa, Ethiopia.

Elias, E., Morse, S., & Belshaw, D. G. R. (1998). Nitrogen and phosphorus balances of Kindo Koisha farms in southern Ethiopia. Agriculture, Ecosystems & Environment 71(1), 93-113.

pg. 93

Esilaba, A. O., Nyende, P., Nalukenge, G., Byalebeka, J. B., Delve, R. J., & Ssali, H. (2005). Nutrient flows and nutrient balances for crop and animal production in smallholder farming systems in eastern Uganda. Agriculture, Ecosystems & Environment 109(3): 192-201.

Ethiopia Agricultural Transformation Agency [EATA]. (2014). Annual report 2013/2014: Transforming Agriculture in Ethiopia. Retrieved September 2, 2015 from http://www.ata.gov.et/annualreport.

Eyasu, E. (1997). Soil fertility management and nutrient balance in Kindo koyisha farms: A case study in North Omo, Southern Ethiopia, K. FRP technical pamphlet, 15.

FAO. (1987). ICS-data, Statistical Division, FAO, Rome.

Fantahun, A. (2015, July 30). Ethiopia’s khat earnings in decline. Ethiopia Observer. Retrieved February 12, 2016, from http://www.ethiopiaobserver.com/2015/08/ethiopias-khat-export- earnings-in-decline.

Francom, M.G. (2015). Ethiopia aims to become one of the world’s top 10 sugar producers. USDA Foreign Agricultural Service. Retrieved March 3, 2016 from http://gain.fas.usda.gov/Recent%20GAIN%20Publications/Ethiopia%20Aims%20to%20/ecome %20One%20of%20the%20World%E2%80%99s%20Top%2010%20Sugar%20Producers%20_Addi s%20Ababa_Ethiopia_11-5-2015.pdf.

Gebremedhin, B., Hirpa, A., & Berhe, K. (2009). Feed marketing in Ethiopia: Results of rapid market appraisal (No. 15). International Livestock Research Institute.

Ghirotti, M. (1998). The role of livestock in mitigating land degradation, poverty and child malnutrition in mixed farming systems: the case of coffee-growing midlands of Sidama, Ethiopia. Livestock and the Environment – International Conference: Animal Production and Health Division by Food and Agriculture Organization of the United Nations (FAO).

Gourley, C.J.P., Powell, J.M., Dougherty, W.J. & Weaver, D.M. (2007). Nutrient balancing as an approach to improving nutrient management on Australian dairy farms. Australian Journal of Experimental Agriculture 47: 1064 -1074.

Haileslassie, A., Priess, J. A., Veldkamp, E., & Lesschen, J. P. (2006). Smallholders’ soil fertility management in the Central Highlands of Ethiopia: implications for nutrient stocks, balances and sustainability of agroecosystems. Nutrient Cycling in Agroecosystems 75(1-3), 135-146.

Hawassa University Agricultural College Soil Laboratory. (2015). Hawassa, Ethiopia.

Haque, T. (2006). Resource use efficiency in Indian agriculture. Indian Journal of Agriculture Economics 61(1): 65-76.

pg. 94

He, C.E., Wang, X., Lui, X., Fangmeier, A., Christie, P., & Zhang, F. Nitrogen deposition and its contribution to nutrient inputs to intensively managed agricultural ecosystems. Ecology Applied 20(1): 80-90.

Heuzé V., Tran G., Baumont R., Noblet J., Renaudeau D., Lessire M., Lebas F. (2015). Wheat bran. Feedipedia, a programme by INRA, CIRAD, AFZ and FAO. Retrieved October 4, 2015 from http://www.feedipedia.org/node/726.

Heuzé V., Tran G., Archimède H., Lebas F. (2015). Sugarcane tops. Feedipedia, a programme by INRA, CIRAD, AFZ and FAO. Retrieved October 4, 2015 from http://feedipedia.org/node/558.

Jackson-Smith, D. (2010). Toward Sustainable Agricultural Systems in the 21st Century. National Academies Press.

Kelly, V.A., & Naseem, A. (2004). Fertilizer use in Sub-Saharan Africa: Types and amounts. Agricultural Sciences 2: 1-7.

Kebede, M. (2012). Determination of Soil P for Optimum Durum Wheat Production in Ada’a, Akaki and Chefe Donsa of The East Shewa, Oromiya Regional State In Central Ethiopia. Journal of Environment and Earth Science 2(1): 1-9.

Kimani, S.K., & Lekasi, J.K. (2004). Managing manures throughout their production cycle enhances their usefulness as fertilisers: A review. In: Bationo A. (Ed.) Managing Nutrient Cycles to sustain Soil Fertility in Sub-Saharan Africa. CIAT-TSBF, Nairobi, Kenya.

Kippie, K.T. (2002). 5000 years of sustainability? A case study on Gedeo land use. PhD Dissertation, Wageningen University, The Netherlands.

Kiros, G., Haile, M., & Gebresamuel, G. (2012). Resources flow and nutrients balance analysis: Nutrients flow associated with the resources and their impact on soil fertility management. PhD Dissertation, Aksum University – Shire Campus. ISBN 978-3-84-84-9928-1: 1-69.

Kiros, G., Haile, M., & Gebresamuel, G. (2014). Assessing the Input and Output Flows and Nutrients Balance Analysis at Catchment Level in Northern Ethiopia. Journal of Soil Science and Environmental Management 5(1), 1-12.

Kumar, B.M., & Nair, P.R. (2004). The enigma of tropical homegardens. In New vistas in agroforestry. Springer Netherlands: 132-152.

Lekasi, J.K., Tanner, J.C., Kimani, S.K., Harris, P.J.C. (2003). Cattle manure quality in Maragua District, Central Kenya: effect of management practices and development of simple methods of assessment. Agriculture, Ecosystems and Environment 94: 289-298.

pg. 95

Leng, R. A., & Preston, T. R. (1976). Sugar cane for cattle production, present constraints, perspectives and research priorities. Tropical Animal Production 1: 1-22.

Lesschen, J. P., Stoorvogel, J. J., Smaling, E. M. A., Heuvelink, G. B. M., & Veldkamp, A. (2007). A spatially explicit methodology to quantify soil nutrient balances and their uncertainties at the national level. Nutrient Cycling in Agroecosystems 78(2): 111-131.

Mahala, A.G., Mokhtar, A.M.S., Amasiab, E.O. & AttaElmnan, B.A. (2013). Sugarcane tops as animal feed. International Research Journal of Agriculture Science and Soil Science 3(4): 147- 151.

Makken, F. (1993). Nutrient supply and distribution at country level case study of Malawi and Ethiopia. Dutch Association of Fertilizer Producers (VPK), Leidschendam: 165-223. Mellisse, B.T., Van de Ven, G., & Descheemaeker, K. (in prep). Chapter One: Home garden system dynamics in Southern Ethiopia. PhD Dissertation, Wageningen University, The Netherlands.

Ministry of Agriculture (Ethiopia). (2000). Agro-ecological Zonations of Ethiopia. Addis Ababa, Ethiopia.

Mitchell, C.C. (2008). Nutrient content of fertilizer materials. Alabama Cooperative Extension System: Auburn University.

Munters, P. J. A. L. (1997). The Dutch Manure Policy: MINAS (Nutrient accounting system) Report from Dutch Dept. of Agriculture of the Ministry of Agriculture, Nature Management and Fisheries. For the Danish Folketing.

Murphy, H. F. (1963). Fertility and other data on some Ethiopian soils. Imperial Ethiopian College of Agriculture and Mechanical Arts.

Mwangi, W. M. (1996). Low use of fertilizers and low productivity in sub-Saharan Africa. Nutrient Cycling in Agroecosystems 47(2), 135-147.

Myburgh, J., Osthoff, G., Hugo, A., De Wit, M., Nel, K., & Fourie, D. (2012). Comparision of the milk composition of free-ranging indigenous African cattle breeds. South African Journal of Animal Science 42(1): 2-14.

Öborn, I., Edwards, A. C., Witter, E., Oenema, O., Ivarsson, K., Withers, P. J. A., ... & Stinzing, A. R. (2003). Element balances as a tool for sustainable nutrient management: a critical appraisal of their merits and limitations within an agronomic and environmental context. European Journal of Agronomy 20(1): 211-225.

pg. 96

Oenema, O., Kros, H., & de Vries, W. (2003). Approaches and uncertainties in nutrient budgets: implications for nutrient management and environmental policies. European Journal of Agronomy 20(1), 3-16.

Paul, S., Onduru, D., Wouters, B., Gachimbi, L., Zake, J., & Ebanyat, P. (2009). Cattle manure management in East Africa. Assessment 94: 289-298.

Roe, M., Pinchen, H., Church, S., & Finglas, P. (2012). Nutrient analysis of eggs: analytical report. Institute of Food Research UK Department of Health: 2-44.

Rotz, C.A. (2004). Management to reduce nitrogen losses in animal production. Journal of Animal Science 82(13): 119-137.

Roy, R.N., Misra, R.V., Lesschen, J.P. & Smaling, E.M.A. (2003). Assessment of soil nutrient balance: Approaches and methodologies. FAO Fertilizer and Plant Nutrition Bulletin 14. Food and Agriculture Organization, Rome.

Smaling, E.M.A. (1993). Soil nutrient depletion in sub-Saharan Africa. In: Van Reuler, H., Prins, W.H. (Eds.), The Role of Plant Nutrients for Sustainable Food Crop Production in Sub-Saharan Africa. Dutch Association of Fertilizer Producers (VKP), Leidschendam: 232.

Spring, A., Hiebsch, C., McCabe, J. T., Tabogie, E., Diro, M., Wolde-Michael, G., & Tesfaye, S. (1997). The Tree Against Hunger: Enset-Based Agricultural System in Ethiopia. Washington, DC, USA: American Association for the Advancement of Science.

Stoorvogel, J. J., & Smaling, E. M. A. (1990). Assessment of soil nutrient depletion in Sub-Saharan Africa: 1983-2000 (Vol. 1). Wageningen: Winand Staring Centre.

Stoorvogel, J.J., Smaling, E.M.A., & Janssen, B.H. (1993). Calculating soil nutrient balances in Africa at different scales: I-supra-national scale. Fertilizer Research 35: 227-235.

Syers, J.K. (1997). Managing soils for long-term productivity. Philosophical Transactions Royal Society of London (B) 352: 1011-1021.

Tadele, R. (2008). Farmer’s Perception of Environmental Degradation and Their Response to Environmental Management: A Case of Dale Woreda, Sidama Zone, SNNPR.

Tena, W. & Beyene, S. (2011). Identification of growth limiting nutrient(s) in alfisols: Soil physico-chemical properties, nutrient concentrations and biomass yield of maize. American Journal of Plant Nutrition and Fertilization Technology 1: 23-35.

The National Agricultural Library. (2015). United States Department of Agriculture. National Nutrient Database for Standard Reference Release 28.

pg. 97

Tizale, C. Y. (2007). The dynamics of soil degradation and incentives for optimal management in the Central Highlands of Ethiopia. PhD Dissertation, University of Pretoria, South Africa.

Tsegaye, Y. (2001). Coffee-Ensete-Livestock interaction for sustainable livelihood in the Sidama area of southern Ethiopia. Journal of Agriculture in the Tropics and Subtropics 102(2): 119-131.

United States Agency for International Development [USAID]. (2005). Southern Nations, Nationalities, and Peoples Region (SNNPR) Overview of Livelihood Profiles. USAID Disaster Prevention and Preparedness Commission (DPPC), The Government of Ethiopia. Retrieved October 14, 2015 from http://pdf.usaid.gov/pdf_docs/Pnadj867.pdf.

Vanlauwe, B., & Giller, K. E. (2006). Popular myths around soil fertility management in sub- Saharan Africa. Agriculture, Ecosystems & Environment 116(1), 34-46.

Van den Broek, J. & Van Hofwegen, G. (2013). Increasing nutrient use efficiency in Dutch agriculture. Retrieved September 20, 2015 from http://www.pbl.nl/sites/default/files/ cms/publicaties/van_den_broek_-_verhoging_n-benutting.pdf.

Van Ittersum, M.K., Claasen, G.D.H., Rossing, W.A.H., Schipper, R.A., Descheemaeker, K., Stoorvogel, J.J., & De Ridder, N. (2014). Quantitative Analysis of Land Use Systems QUALUS. Wageningen University PPS-30306 Course Guide: Plant Production Systems, Operational Research and Logistics, Farming Systems Ecology, Development Economics, Soil Geography and Landscape.

Van Heerden, S.M., Schonfeldt, H.C., Smith, M.F., & Jansen van Rensburg, D.M. (2002). Nutrient content of South African Chickens. Journal of Food Composition and Analysis 15: 47-64.

Wallace, M.B., & Knausenberger, W.I. (1997). Inorganic fertilizer use in Africa: Environmental and economic distribution. Environmental and Natural Resources Policy and Training (EPAT) Project: Winrock International Environmental Alliance: 1-53.

Wander, M. (2015). Nutrient balance basics for organic farming. eOrganic. Retrieved March 3, 2016 from http://articles.extension.org/pages/18794/nutrient-balance-basics-for-organic- farming-systems.

Ward Jr, J.H. (1963). Hierarchical grouping to optimize an objective function. Journal of the American Statistical Association 58(301), 236-244.

Wondo Genet College Soil Laboratory. (2015). Wondo Genet, Ethiopia.

pg. 98

7. APPENDICES

7.1 CONVERSION TABLE

Unit in kilogram(s) Chinet 50.00 Cup (milk) 0.25 Egg 0.06 Enset leaf 1.93 Esir 1.00 Gimbola 9.78 Kutal 100.00 Liter (milk) 1.00 Shekim 12.48 Zurba 1.00

pg. 99

7.2 NUTRIENT CONTENT

Crop type Specification Dry matter (%) N (ppm) Total N (%) P (ppm) Total P (%) K (ppm) Total K (%) Reference (N) Reference (P & K) 7 year 31.15 4110 0.41 1301 0.13 3150 0.32 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 6 year 7520 0.75 1455 0.15 5175 0.52 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 5 year 8460 0.85 1828 0.18 9975 1.00 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Enset kocho 4 year 18330 1.83 1394 0.14 7200 0.72 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 3 year 18800 1.88 1642 0.16 6075 0.61 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Average 11444 1.14 1524 0.15 6315 0.63 7 year 53.69 9870 0.99 1828 0.18 3650 0.37 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 6 year 9400 0.94 3321 0.33 6175 0.62 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Enset bula 5 year 10340 1.03 3057 0.31 4050 0.41 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Average 9870 0.99 2736 0.27 4625 0.46 7 year 12.85 13630 1.36 5187 0.52 44975 4.50 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 6 year 14.69 10340 1.03 4440 0.44 53000 5.30 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 5 year 15.11 16450 1.65 5560 0.56 45275 4.53 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Enset leaf 4 year 13.30 12220 1.22 3321 0.33 39625 3.96 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 3 year 12.76 13160 1.32 3881 0.39 47325 4.73 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Average 13.74 13160 1.32 4478 0.45 46040 4.60 From khat, enset Grass 33.00 16300 1.63 4900 0.49 19600 1.96 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) and coffee fields Coffee berry None 36.00 15510 1.55 3134 0.31 31875 3.19 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Coffee bean None 36.04 22090 2.21 4627 0.46 21625 2.16 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Grain 78.13 11750 1.18 6679 0.67 10200 1.02 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Grain 78.13 13160 1.32 7052 0.71 8200 0.82 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Grain average 78.13 12455 1.25 6866 0.69 9200 0.92 Barley Straw 71.24 7050 0.71 3694 0.37 2490 0.25 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Straw 71.24 5170 0.52 2388 0.24 14175 1.42 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Straw average 71.24 6110 0.61 3041 0.30 8333 0.83 Grain 70.27 11280 1.13 5373 0.54 5075 0.51 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Maize Stover 33.10 6580 0.66 16731 1.67 14175 1.42 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Cabbage None 25.00 20800 2.08 26000 2.60 16960 1.70 The National Agricultural Library (2015) The National Agricultural Library (2015) Leaf 22.76 22090 2.21 5000 0.50 26750 2.68 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Onion Root 22.76 49350 4.94 5000 0.50 24175 2.42 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Average 22.76 35720 3.57 5000 0.50 25463 2.55 Dwarf leaf 33.79 11800 1.18 3838 0.38 13250 1.33 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Dwarf twigs 25.04 17400 1.74 5933 0.59 11125 1.11 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Khat Tall leaf 34.40 14100 1.41 4254 0.43 18200 1.82 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Tall twigs 25.71 14100 1.41 4627 0.46 19750 1.98 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Average 29.74 14350 1.44 4663 0.47 15581 1.56 Milk None n/a 3610 0.36 930 0.09 1550 0.16 Myburgh et al. (2012) Myburgh et al. (2012) Same as milk (for this Butter n/a 3610 0.36 930 0.09 1550 0.16 Myburgh et al. (2012) Myburgh et al. (2012) study) Eggs None n/a 2016 0.20 1450 0.15 1790 0.18 Roe et al. (2012) Roe et al. (2012) Both male and female, minerals for raw chicken meat obtained Chicken n/a 84000 8.40 5380 0.54 6820 0.68 Van Heerden et al. (2002) Van Heerden et al. (2002) from skin, white and dark meat (1.3 kg body weight assumed) pg. 100 Small ruminants Male (30 kg empty body n/a 25000 2.50 6000 0.60 2000 0.20 Agricultural Research Council (1984) Agricultural Research Council (1984) (goat, sheep) weight assumed) Small ruminants Female (30 kg empty n/a 23000 2.30 6000 0.60 2000 0.20 Agricultural Research Council (1984) Agricultural Research Council (1984) (goat, sheep) body weight assumed)

Both male and female Large ruminant (500 kg typical mature n/a 24320 2.43 7233 0.72 1940 0.19 Agricultural Research Council (1984) Agricultural Research Council (1984) (cattle) cow weight assumed) Average is taken. Homegarden Includes manure and 26.37 8300 0.83 300 0.03 2900 0.29 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) compost household leftovers Crop type Specification Dry matter (%) N (ppm) Total N (%) P (ppm) Total P (%) K (ppm) Total K (%) Reference (N) Reference (P & K) 7 year 31.15 4110 0.41 1301 0.13 3150 0.32 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 6 year 7520 0.75 1455 0.15 5175 0.52 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 5 year 8460 0.85 1828 0.18 9975 1.00 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Enset kocho 4 year 18330 1.83 1394 0.14 7200 0.72 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 3 year 18800 1.88 1642 0.16 6075 0.61 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Average 11444 1.14 1524 0.15 6315 0.63 7 year 53.69 9870 0.99 1828 0.18 3650 0.37 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 6 year 9400 0.94 3321 0.33 6175 0.62 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Enset bula 5 year 10340 1.03 3057 0.31 4050 0.41 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Average 9870 0.99 2736 0.27 4625 0.46 7 year 12.85 13630 1.36 5187 0.52 44975 4.50 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 6 year 14.69 10340 1.03 4440 0.44 53000 5.30 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 5 year 15.11 16450 1.65 5560 0.56 45275 4.53 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Enset leaf 4 year 13.30 12220 1.22 3321 0.33 39625 3.96 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) 3 year 12.76 13160 1.32 3881 0.39 47325 4.73 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Average 13.74 13160 1.32 4478 0.45 46040 4.60 From khat, enset Grass 33.00 16300 1.63 4900 0.49 19600 1.96 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) and coffee fields Coffee berry None 36.00 15510 1.55 3134 0.31 31875 3.19 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Coffee bean None 36.04 22090 2.21 4627 0.46 21625 2.16 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Grain 78.13 11750 1.18 6679 0.67 10200 1.02 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Grain 78.13 13160 1.32 7052 0.71 8200 0.82 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Grain average 78.13 12455 1.25 6866 0.69 9200 0.92 Barley Straw 71.24 7050 0.71 3694 0.37 2490 0.25 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Straw 71.24 5170 0.52 2388 0.24 14175 1.42 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Straw average 71.24 6110 0.61 3041 0.30 8333 0.83 Grain 70.27 11280 1.13 5373 0.54 5075 0.51 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Maize Stover 33.10 6580 0.66 16731 1.67 14175 1.42 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Cabbage None 25.00 20800 2.08 26000 2.60 16960 1.70 The National Agricultural Library (2015) The National Agricultural Library (2015) Leaf 22.76 22090 2.21 5000 0.50 26750 2.68 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Onion Root 22.76 49350 4.94 5000 0.50 24175 2.42 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Average 22.76 35720 3.57 5000 0.50 25463 2.55 Dwarf leaf 33.79 11800 1.18 3838 0.38 13250 1.33 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Dwarf twigs 25.04 17400 1.74 5933 0.59 11125 1.11 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Khat Tall leaf 34.40 14100 1.41 4254 0.43 18200 1.82 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Tall twigs 25.71 14100 1.41 4627 0.46 19750 1.98 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) Average 29.74 14350 1.44 4663 0.47 15581 1.56 Milk None n/a 3610 0.36 930 0.09 1550 0.16 Myburgh et al. (2012) Myburgh et al. (2012) Same as milk (for this Butter n/a 3610 0.36 930 0.09 1550 0.16 Myburgh et al. (2012) Myburgh et al. (2012) study) Eggs None n/a 2016 0.20 1450 0.15 1790 0.18 Roe et al. (2012) Roe et al. (2012) Both male and female, minerals for raw chicken meat obtained Chicken n/a 84000 8.40 5380 0.54 6820 0.68 Van Heerden et al. (2002) Van Heerden et al. (2002) from skin, white and dark meat (1.3 kg body weight assumed) Small ruminants Male (30 kg empty body n/a 25000 2.50 6000 0.60 2000 0.20 Agricultural Research Council (1984) Agricultural Research Council (1984) (goat, sheep) weight assumed)

Small ruminants Female (30 kg empty n/a 23000 2.30 6000 0.60 2000 0.20 Agricultural Research Council (1984) Agricultural Research Council (1984) (goat, sheep) body weight assumed)

Both male and female Large ruminant (500 kg typical mature n/a 24320 2.43 7233 0.72 1940 0.19 Agricultural Research Council (1984) Agricultural Research Council (1984) (cattle) cow weight assumed) Average is taken. Homegarden Includes manure and 26.37 8300 0.83 300 0.03 2900 0.29 Hawassa Soil Lab (2015) Wondo Genet Soil Lab (2015) compost household leftovers

pg. 101

7.3 SURVEY: INPUTS AND OUTPUTS IN FIVE HOMEGARDEN TYPES IN SOUTHERN ETHIOPIA

Name of household head: Sex : Woreda (District): Kebele (Village): Agro-ecology: Altitude: Distance from Hawassa (km): Distance from major roads: Wealth status: Coordinates:

Form 1: Household composition

Main occupation Main occupation Presence (days HHM# Name Gender Age Relation to HH Education level (1) (2) per month) HHM 01 HHM 02 HHM 03 HHM 04 HHM 05 HHM 06 HHM 07 HHM 08 HHM 09 HHM 10 HHM11 HHM12 HHM = household member

pg. 102

Form 2: Description of fields (F) from September 2014—August 2015

Distance Crop type in Area from HS specific field (m2/ha) Input to different fields (m) Fertilizer Manure/compost Other Price DAP Urea When Herbicide When Price When Price When Quantity Unit per (kg) (kg) (month) (L/kg) (month) unit Front yard grazing Enset Coffee Coffee +Enset Khat Maize Barley Wheat Teff Vegetables Root and tuber

Other Seed Price Malatine Price per When DDT Price When (month) Kilogram (per (L/kg) unit (ዋጋ) (month) (L/kg) (ዋጋ) kg) Khat Maize Barely Wheat

Checklist: seed, fertilizer (e.g. DAP, urea), manure (e.g. FYM cattle, FYM chicken), pesticides (e.g. fungicide, herbicide, insecticide), hired labor (weeding for specific F), machinery rent, cropping aids (sticks, plastic), fuel for irrigation and more.

pg. 103

Form 2-I: Crop management (inputs for each specific field)

Nursery (seed bed preparation) Transplanting /planting Weeding Harvesting /processing When No Price per When No Price When No Price When No labour Price /day Date/Month labour unit/day Date/Month labour /day Date/Month labour /day Date/Month

Enset

Coffee

Khat

Maize

Barley

Wheat

Teff

Vegetables

Root and tuber

pg. 104

Form 2-O: Crop yields & residue management (outputs from each field)

No enset harvested/crop yield Enset leave sold When Amount (kg/local measurement) Amount sold (kg/local chinet) No Frequency Date/Month Kocho Bula Kacha Kocho Price Bula Price Amount leave Price

Enset

Fresh Dry Fresh Price Dry Price Coffee

Harvest Time of harvesting/year Harvest 2 Harvest 3 Harvest 1 Price Harvest 2 Price Harvest 2 Price 1 1. September, 2. March, 3. June-July No zurba No zurba No zurba Khat Vegetable

s Sugar cane Maize Barley Wheat Teff

Checklist: outputs to anywhere; crop products for sale (to EXT), for consumption/storage (to SA), for processing (to OA); residues for animals (to AA) and composting.

pg. 105

Form 2A: Animal numbers and changes

Now (2015) 2014/2015 Sold (-) Born (+) Died (-) Consumed (-) Other in (+) Other out (-) AA (sub)type No. No No. No. No. No. No. No. Lactating Cow Dry cow Oxen castrated Oxen intact (bull) Heifer Calves Sheep adult male Sheep young male Sheep adult female Sheep young female Goat adult male Goat young male Goat adult female Goat young female Donkey Horse Chicken Bees traditional hive Bees modern hive

pg. 106

Form 2A-I: Animal feeding & care (inputs in animal activities)

Name of product Grass from your farm Enset leave Concentrate Others Labour (shekim) Communal Dry season Rainy season Rainy season Dry season Rainy season Dry Rainy land No. enset Furushka Furushka No. months No. enset (leaf/day) Enset Coffee Khat Price Price No No (leaf/day) (kg) (kg) /year Cattle cross bred Cattle indigenous Sheep Goat Donkey Horse Chicken Tree twigs and leave Sugarcane top Fodder Salt Medicine Dry season Rainy season Dry Rainy Dry Rainy Dry Price Rainy Price Cost (birr) Cattle cross bred Cattle indigenous Sheep Goat Donkey Horse Chicken

Checklist: inputs from own farm: enset leaves, tree twigs and leaves, sugarcane top, fodder. Purchased: purchased feeds (e.g. concentrates, crop residues, salt lick, sugarcane top), veterinary services.

pg. 107

Form 2-O: AA: Animal production (output from animal activities)

Milk/ butter or egg produced (kg/liter) Sold Livestock type Milk/day Butter Egg/year Milk (kg)/day Price When Butter Price When Egg Price Cattle cross bred Cattle indigenous Sheep Goat Chicken

Checklist: outputs to anywhere; animal products (milk, eggs, skins) for sale (to EXT) or for consumption (to SA); manure and live animal for sale.

Form 3-I: Manure & compost inputs (inputs in redistribution activities or to compost pit)

To compost pit To fields directly Enset field Coffee field Khat field When (every day Cow When (every day or Cow When (every day Cow dung Household waste Cow dung or weekly dung weekly dung or weekly Cattle Sheep Goat Donkey Horse Chicken

Checklist: inputs from outside purchased manure and compost, residues for composting, compost starters and enrichments or own farm.

pg. 108

Form 3-O: Manure & compost management (outputs from redistribution activities or from compost pit)

From compost to... When (e.g. daily, weekly, monthly, Quantity per application Unit Reason (sale or 2 times a year etc. time fields) Enset field Coffee field Enset and coffee Khat field Maize field Barley field Vegetable field

Assets of the farms No Price Cart Motorbike Car Mill

Checklist: outputs to anywhere; compost and manure for use on crops (to F) or sales (to EXT).

pg. 109

Form 4-I: Services obtained at household level (labour hire in)

To When What No Date/Month Name of service/activities Quantity (no) Unit (man day or---- Price per unit Remarks 1 Khat field 2 Enset field 3 4 5 6 7

Ext= external checklist.

Form 4-O: Off-farm labor (services provided by household members or labour hire out)

HHM When What Responsible No. Date/Month Name of service HHM# Quantity Unit (days) Price per unit Remarks 1 Father 2 Son 3 Daughter 4 5

HHM; services provided by household members; off-farm labor (preferably recorded in no. of days), rent received from land rented out.

pg. 110

Form 5-I: Inputs into stock (external inputs into storage activities or purchased items)

No When What Date/Month Name of product Quantity Unit Price per unit Remarks 1 2 3 4 5

Checklist: inputs from outside; purchases of staple food (grains and pulses)

Form 5-O: Outputs from stock (output from storage activities)

No When What Remarks (sale or Date/Month Name of product Quantity Unit Price per unit sowing)

Checklist: outputs to anywhere; use of products in stock for sowing, sales (to EXT), and Household consumption.

pg. 111